Vous êtes sur la page 1sur 28

Conjunctive cohesion in four English genres

RAOUL N. SMITH and WILLIAM J. FRAWLEY

Abstract

A major ingredient of textuality is cohesion. A text is not a text unless it


coheres. But different text types do not cohere in the same way. In thispaper,
we foats on one type ofcohesive tie, conjuction, and compare its use in four
different American English genres — fiction, journalism, religion, and science.
Our results show that methods of conjunction in these genres vary in a statisti-
cally significant way and that conjunctions, althoughfew in number of types
and tokens, play a major wie in structuring these different text types.

l. Introduction

If one looks in some detail at the various models which have been proposed
for text structure, one finds that text researchers have been struggling, either
explicitly or implicitly, with the problem of the Connectivity oftext since it
appears that Connectivity is the one thing that makes a text a text. Although
text linguists have pursued Connectivity rather intensely, they have not always
met with a great deal of success in characterizing or defining how text holds
together.
Grimes' (1975) much-used model of text has provided a fairly interesting
approach to textual Connectivity in terms of rhetorical predicates, or high-
level rhetorical connectives which organize the network of propositions which
constitute texts. However, while Grimes' model has been used with some suc-
cess (e.g. Meyer, 1975; Frawley, 1980), rhetorical predicates remain rather
unwieldy and ill-circumscribed notions. One has only to look at the difficulty
which Meyer had in applying Grimes' notions to her rather simple textual
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
0165-4888/83/0003-0347 $2.00 Text 3(4)Authenticated
(1983), pp. 347-374
© Mouton Publishers, Amsterdam Download Date | 9/16/19 12:44 PM
348 RaoulN. Smith and William J. Frawley

Stimuli: Grünes provides few criteria for deciding when and where a text is
connected by a certain rhetorical predicate.
Other research which has focused on Connectivity has been less successful
or generalizable than Grimes'. Schank and Abelson's (1977) reduction oftext
to sequences of atomic propositions relies heavily on a Causal Syntax for
Connectivity. The AI System which has come out of this study works well äs
an AI System, but the notion of Connectivity which they propose is unnatural:
for Schank and Abelson, all propositions in texts are connected causally, and
this is decidedly not the case in texts outside their AI System, most of which
have very little causal Connectivity (äs we show subsequently).
Similarly, Petöfi's (1975) and van Dijk's (1977a and b) studies oftext logic
have revealed that intratextual Connectivity can be characterized and formal-
ized by a first order (and sometimes more complex) logic, but while it is
valuable to demonstrate that text logic exists, it is another thing to demon-
strate what kind of logic text logic is. The translation of textual connectives
(äs the translation of all natural language connectives) into the connectives of
first order logic entails the blurring of subtle distinctions in natural language
for the sake of formalizability. For example, in natural language text, it is
possible to have various types of sequential connection - spatial, temporal,
hypothetical, and causal — but these would all be collapsed into a simple logi-
cal implication and/or logical conjunction. This collapsing may be good for
formal Systems, but it is detrimental to the discovery of exactly how texts
cohere: the translation of natural language connectives into logical ones elim-
inates the interesting types of Connectivity possible in natural language.
It seems to us that the only theoretical approach to textual Connectivity
which has enjoyed a modicum of success has been Halliday and Hasan's(1976)
work on cohesion. Their detailed work on textual connection, though oc-
casionally suspect in its categorization, has at least provided a starting point
for a systematic inventory of types of textual connection while avoiding
Grünes' problem of unwieldiness, Schank and Abelson's reduction of connec-
tion to causality, and Petöfi's and van Dijk's translation of Connectivity into
logical Connectivity.
Halliday and Hasan's scheme has been used rather successfully in the past
(see, e.g. Wise and Stewart, 1981; and Frawley and Smith, 1983), and we
propose to further their work by doing a systematic computational analysis of
patterns of conjunctive cohesion in four genres of English text. We pose this
problem of cross-genre textual Connectivity for two reasons. First, conjunc-
tion is a critical device for the coherence of a text. Conjunctions are relational
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 349

entities; they are like textual prepositions in that they Signal relationships of
textual constituents to each other on a textual level. This analogy is perhaps
clearer through example. The function of prepositions is to mark, on the intra-
clause level, the relationships of clausal constituents to each other. The scope
of prepositions never exceeds clause boundaries, but serves only to link nomi-
nals with predicates; in this sense, prepositions are actually intra-clausal con-
junctions. Conjunctions, however, always point outside the clause itself. They
link two or more overt clauses, äs in 'After John came, I left', where 'after'
Signals a sequential relationship between twopropositions; or they link two or
more implicit clauses, äs in 'John and I are here', where 'and' Signals the pres-
ence of an implicit second clause which has been reduced by transformation
and collapsed into a coordinate subject. Conjunctions mark the cross-clausal
relationship of the semantics of clauses to each other: in this way, conjunc-
tions are indicators of what might be called 'textual case' (whereas preposi-
tions mark 'clausal case'). Thus, insofar äs the Connectivity of textual constit-
uents is critical to defining what makes a text a text, Conjunctions, äs textual
prepositions, merit study because of their cross-causal function.
The second reason for posing the above problem deals with the cross-genre
focus of our study. Despite the fact that conjunction has been studied in
some detail in the past, most of the work has been done on conjunction in
narrative, äs, similiarly, the legacy of 'story grammar' has been studied in rela-
tion to linguistics. If conjunction is so crucial a textual device äs we make it
out to be, then it ought to be manifested differently in different modes of
texts (genres). Some evidence already exists that different genres use other
textual devices differently (see Frawley and Smith, 1983), and if conjunction
contributes to Connectivity and if Connectivity is what makes a text a text,
then it is reasonable to assume (beyond the simple Intuition that there are
different genres) that different modes of text will connect in different ways.
Thus, we are motivated to look at conjunction not only because of its intrinsi-
cally interesting properties, but also because its function may be made clearer
through its differential distribution across genres.
To this end, we have studied 16,000 word samples of fiction, journalism,
religion, and science from the Brown English Corpus. We have completed fre-
quency distribution studies both lexically and syntactically on the corpora,
and we are reporting on what we have discovered thus far about the differen-
tial distribution in the genres of conjunction. In the data and discussion be-
low, we are striving to report on not only the different (and similar) patterns
of conjunction in the genres, but also on the discursive and epistemological
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
350 RaoulN. Smith and William J. Frawley

implications of the patterns. It is our belief that conjunction and all textual
connectives provide an insight into the logic of the discourses (see Foucault,
1972) which we are investigating. Thus, patterns of textual conjunction reveal
the fundamental conditions on the connection of ideas in the genres in ques-
tion, and we hope to explain what makes these genres what they are in terms
of their respective practices of Connectivity.

2. The distribution of conjunctions in general:


Some preliminary problems

Let us begin this study of different cohesive conjunction patterns with a look
at the general distribution of conjunctions (both subordinate and coordinate)
according to genre. This will give us a preliminary feel for the patterns we are
attempting to explain.
The frequency of occurrence of coordinating and subordinating conjunc-
tions in the four genres appears in Table l. Although there is a wide variety in
the frequencies of conjunctions among the genres, especially the coordinating
ones, an examination of the proportions of each conjunction category to the
total reveals that the genres are quite different. Coordinating conjunctions
constitute 66.4% of the conjunctions in fiction, 54.7% ofthose in journalism,
55.4% in religion, and 59.4% in science. (Of course, the proportions of sub-
ordinating conjunctions are complements of these, namely 33.6%, 45.3%,
44.6% and 40.6%, respectively). The ranges, and therefore the ratios, vary
widely among genres. In fact, statistical tests of the non-proportional frequen-
cies of coordinating vs. subordinating conjunctions show significant differ-
ences for use of coordination over Subordination for all genres (X2 = 35.006,
p < .001). This is a rather surprising finding, given what recent measurements
of syntactic maturity have suggested about written language. Sophisticated
prose is supposed to be highly subordinate (e.g. Hunt, 1964), but these
data clearly show that even in sophisticated prose, coordination is much more
prevalent than Subordination, which must say something about the natural-
ness of coordination over Subordination in linguistic performance, no matter
what the level of sophistication (Givon, 1979, makes similar claims about
discourse).
However, a closer inspection of the data reveals that some genres are more
conjunctive than others. In particular, religion and fiction are more conjunc-
tive than science and journalism. But within each genre, there appear to be
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 351

Table l. Conjunctions in four genres

Coordinating Conjunctions
Conjunction Fiction Journalism Religion Science

also 0 (Dl 1 0
and (33)537 (5)300 (25)445 368
but (41)92 (23)48 (20)83 12
nor 2 0 (3)14 1
or (3)53 38 (2)79 57
plus 0 2 0 1
yet (4)8 0 (9)12 0

Totais: (81)692 (29)389 (59)634 (0)439

Subordinating Conjunctions
Conjunction Fiction Journalism Religion Science

after (2)7 (2)13 2 (1)2


although (2)2 (1)3 (5)8 (5)11
äs (5)88 83 (10)150 (5)92
because (1)10 5 (6)19 5
before (1)8 (1)4 (1)5 5
for (6)13 5 (17)22 6
if (13)50 (5)21 (12)41 (10)21
lest 4 0 0 0
like (2)45 4 8 2
since (4)10 7 (2)6 (4)16
so 4 4 (1)12 9
than 13 10 21 40
that 96 138 166 87
though 0 1 18 4
unless 0 2 0 0
until 0 (1)3 4 0
whereas 0 0 2 0
whereupon 0 0 (1)2 0
whether 0 9 11 0
while 0 (4)10 (1)12 0

Totais: (36)350 (14)322 (56)509 (25)300

Total Conjunctions: 1042 711 1143 739

Mean
sentence length: 14.4 21.4 23.2 23.8

(The numbers in parentheses indicate the cohesive uses of these Conjunctions.)

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
352 RaoulN. Smith and William J. Frawley

trends toward a preference for one type of conjunction over another. In par-
ticular, journalism and science use coordination much less frequently than
fiction and religion, but in fiction, for example, coordination is twice äs fre-
quent äs Subordination, but in journalism they are almost equi-frequent. These
are peculiar findings, and we suggest, in the subsequent detailed analysis, why
these differences occur.
Another peculiarity can be seen in a comparison of frequency of conjunc-
tion to mean sentence length. One would expect that there would be a corre-
lation of the frequency of conjunction with sentence length. In particular, it
is reasonable to surmise that genres which use a high number of conjunctions
would also have a high mean sentence length. But this is the case only with
religion. In the million word corpus the journalism and science texts, which
have relatively fewer conjunctions, have a relatively long mean sentence length:
21.4 and 23.8 words per sentence respectively (the mean for the whole corpus
in 19.3). On the other hand, fiction, which uses a great number of conjunc-
tions, has a very short mean sentence length (14.4) compared to the other
three genres. These findings, together with the findings about the distribution
of conjunctions in general in the genres, suggest that there are distinct patterns
of connection across these text types. Some answers to these puzzles will be
proposed below.

3. Coordinating conjunctions

A more indepth look at coordination shows some basic conjunctive patterns.


Seven coordinating conjunctions appear in the texts: also, and, but, nor, or,
plus, and yet. Also, nor, plus, and yet occur infrequently. Nor occurs four-
teen times in the religion text and only three times in the other three genres
combined. Yet occurs twelve times in religion and eight times in fiction, but
not at all in journalism or science. The other three coordinating conjunctions
occur relatively frequently in all four genres, äs can be seen from the follow-
ing table of percentage of tokens:

Fiction Journalism Religion Science

and 77.6% 77.4% 70.2% 83.8%


but 13.3% 12.3% 13.1% 2.7%
or 7.6% 9.7% 12.4% 13.0%
others 1.5% 0.6% 4.3% 0.5%

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 353

An X2 test was applied to all possible combinations of the fiction, journal-


ism, religion and science samples. All genres differed significantly from each
other at the .01 level, except for fiction and journalism, which behave quite
similarly with respect to coordinating conjunctions and represent a sort of
neutral or middle ground between religion and science. But this clustering is
understandable. Fiction and journalism are quite similar generally in that each
involves the creation of a sequential event line: Gutwinski (1976), in fact,
found a predominance of coordinate cohesion in his analysis of connectives in
literary text. Religion and science, of course, differ from fiction and journalism
in this respect since they are more logic-oriented. Greenbaum (1969) has
made this very point: that discourses or argument and exposition (such äs
science and philosophy) demand greater use of logical sentence adjunction
than does sequential narration. That religion is low in coordination is under-
standable from its nature äs hypothetical discourse, but why should science, a
logic-oriented discourse, have the most coordination? Cheong's (1978)ex-
planation is relevant here. Coordination contributes to right branching in syn-
tax, which has long been claimed to be cognitively simpler. Thus, it appears
that the high functional load of sentences in scientific texts, with a predomi-
nance of discourse-specific lexical items, demands a cognitively simple syntax
for processing by a reader (note, in fact, that science has the highest mean
sentence length). Further comparison of the differences between science and
religion shows that science uses 13.6% more of the additive conjunction and
than religion but 10.4% less of the adversative or contrastive but. This is an
indication of the rhetorical structure of science and religion texts: it suggests
a list or sequential argument structure in science, and an assertion and contrast
structure for religion. This latter conclusion is further corroborated by the
high frequencies of nor and yet in religion texts and the substantial inverse
differential usage of or and but in science (57 and 12, respectively) and in
religion (79 and 83, respectively).

4. Subordinating conjunctions

A look at the data on subordinating conjunctions reveals both similarities to,


and differences from, that on coordination. First, the variety of subordinating
conjunction types is much greater than the coordinating ones. There are almost
three times äs many different subordinating conjunction types represented in
the texts than coordinating ones. This is in spite of the fact that, asmentioned
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
354 RaoulN. Smith and William J. Frawley

above, Subordination, äs a textual device, is less frequent. But it is obviously


the case that the coordinating conjunction and has a much higher functional
load than any other conjunction, subordinating or coordinating, so that its
effect, and therefore the effect of coordination on the total token distribution,
is quite strong.
Of the low frequency subordinating conjunctions, it is interesting to note
that whereas and whereupon, of frozen regist er, occur only in religion (twice
each), while until, while and whether (the latter two about ten times each)
occur only in religion and journalism, and unless only in journalism (twice).
(The significance of this for cohesion will be discussed below.) Lest, generally
an infrequent word, (it occurs only seventeen times in the million word cor-
pus in ten of the fifteen genres and thirteen of the 500 samples) occurs four
times in fiction (three times in Guy Endore's Voltaire, Voltaire with period
dialog, and in W.E.B. Dubois' Worlds ofColor).
The two most frequent subordinating conjunctions in all four genres are
that and äs. That is somewhat more frequent than äs in three of the genres —
fiction, journalism, and religion — but the reverse holds true for science. The
less frequent occurrence of that in science may be due to the fact that this
genre tends to be more nominal and less verbal (Smith, ms) and since that is a
frequent verbal complementizer, it would tend to occur less often in science.
Of the remaining eleven conjunctions occurring in the mid-range, it is
interesting to notice which conjunctions appear the most frequently in the
four genres. The highest frequencies for before, ift and like occur in fiction,
which may say something about seriation and analogy in fictive texts. Journal-
ism has the most occurrences ofafter, no doubt also related t o seriation. Reli-
gion has the highest frequencies for because, for, so, and though (the latter
being very infrequent in the other three genres), and their appearance is surely
related to causal and contrastive text structure in the genre. Science uses al-
though, since, and than more frequently than the other genres, which suggests
contrastive and comparative text logic in such texts.
This describes the general distribution of each type of conjunction. So far,
this study has focused on a comparative use of the conjunctions in these four
text types, but it has not yet addressed itself to a comparative study of their
cohesive function. Yet conjunctions do function äs cohesive devices and, in
the light of the problems and patterns heretofore discussed, the rest of this
paper focuses on the cohesive function of conjunction.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 355

5. Cohesive conjunction

The Standard work on cohesion, Halliday and Hasan 1976, proposes a very
narrow definition of conjunctive cohesion. For these t wo researchers, con-
junctive cohesion occurs extra- or cross-sententially and apparently only in
sentence initial position. All other uses, especially phrasal use, are termed
structural. With this restrictive definition we find, äs a matter of fact, that
conjunction äs a whole is not used very often äs a cohesive device in texts.
Within the bounds of their restricted definition, the percentage of conjunc-
tions used cohesively in these genres appears in the following table:

Fiction Journalism Religion Science

Cohesive
conjunction 10.4% 5.9% 7.8% 3.2%

It is clearly the case that Miss Fidditch's admonition to avoid the initial
conjunction has been strongly heeded by scientists, but not by fiction writers,
to the same extent.
This is especially clear when one examines the cohesive coordinating con-
junctions:

Fiction Journalism Religion Science

Cohesive
coordinating
conjunctions 11.0% 7.2% 6.6% 0.0%

Scientists in this sample never used coordinating conjunctions cohesively.


Subordinating conjunctions used in a cohesive manner show a different
pattern, however.

Fiction Journalism Religion Science

Cohesive
subordinating
conjunctions 9.1% 4.3% 9.2% 8.0%

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
356 RaoulN. Smith and William J. Frawley

A comparison of these two tables shows that fiction and journalism use
coordinating conjunctive cohesion more frequently than subordinating cohe-
sion, but in religious and scientific texts the exact opposite holds true. How
are these data to be interpreted? Or, more precisely, what is the function of
conjunctive cohesion and why do these genres differ in their usage? Halliday
and Hasan do not explicitly state the cohesive function of conjunction, but
they do hint at it (sometimes in somewhat contradictory terms).

Conjunctive elements are cohesive not in themselves but indirectly, by virtue


of their specific meanings; they are not primarily devices for reaching into the
preceding (or following) text, but they express certain meanings which pre-
suppose the presence of other components in the discourse (Halliday and
Hasan, p. 226).

But later,

When we are considering these sentences specifically from the point of view
of cohesion . . . we are inevitably concerned with their actual sequence äs
expressed, because cohesion is the relation between sentences in a text, and
the sentences of a text can only follow one after the other. Hence in describing
conjunction äs a cohesive device, we are focusing attention not on the seman-
tic relations äs such . . . but on one particular aspect of them, namely the
function they have of relating to each other linguistic elements that occur in
succession but are not related by other, structural means (Halliday and Hasan,
p. 227).

The genre-specific usage that we see in the data above can be explained, in
the case of fiction and journalism, by the fact that the differences in these
two genres are in direct proportion to the use of these conjunction categories
cohesively and non-cohesively, that is, to their total occurrcnccs. In the casc
of religion and science, however, that Statement cannot be made. The cohesive
use of conjunctions in the latter two genres is significant. Subordinating con-
junctions are used cohesively more often than coordinating conjunctions in
religion and science. The influence of genre style constraints is clearly a factor.
But, in addition to the more easily perceived, Standard, gross differences in
genres, it has become clear that they also structure their texts differently,
even in the types of conjunctions they favor, and in particular, conjunction
choice in genres is semantically conditioned.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 357

5.1. The semantics of cohesive conjunction

Halliday and Hasan classify the conjunctions into four semantic categories —
additives, adversatives, causals, and temporals. It is clear that they do not fol-
low the traditional classification of coordinating vs. subordinating, but that
is because such a classification is syntactically motivated and Halliday and
Hasan's research is centered on semantics. It also shows why sentence initial
conjunction is their principal focus.
However, Halliday and Hasan's definition of the scope of cohesive con-
junction is somewhat narrow. From their examples, but no explicit Statement,
cohesive conjunction is limited to sentence initial position. They do admit,
however, that the determination of sentence boundary is rather abitrary. If
one broadens the definition of conjunctive cohesion beyond traditional sen-
tence boundaries (and in these texts this means including the period and ques-
tion mark) to include semi-colons, the tie is still inter-sentential, but in a wider
sense. The genre most affected by this change is religion which shows sub-
stantial changes in use. In particular, coordinating conjunctions after a semi-
colon (only and, but, and yef) occur twenty times in religion, five times each
in fiction and science (all with and in science) and only once in journalism.
Under this new scope of cohesive conjunction, the total percentage of con-
junctions used cohesively then becomes:

Fiction Journalism Religion Science

Coordinate 12.6% 7.4% 9.8% 1.1%


Subordinate 11.1% 4.3% 11.0% 8.3%

All categories increase except for Subordination in journalism. Religion


then moves into second place and this fact also helps to explain the high
mean sentence length for this genre. (A contrastive study of phrasal vs. sen-
tential conjunction is underway in order to make clearer comparisons between
fiction and religion). It is with this broadened definition of the scope of
cohesion that we will be operating in the rest of this paper. It is the frequen-
cies of this use which appear in Table l in parentheses.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
358 RaoulN. Smith and William J. Frawley

5.2. Cohesive coordination

Adopting Halliday and Hasan's typology of additive, adversative, causal, and


temporal conjunctive relations, we find only two of these classes being used
cohesively in the set of coordinating conjunctions appearing in the four
samples. These are the additive and the· adversative. The additives occurring
cohesively in the texts are also, and, nor (negative additive), and or (alternative
additive). The adversatives are but anayet.
The percentage of additives used cohesively in each genre is:

Fiction Journalism Religion Science

also 0.0% 100.0% 0.0% 0.0%


and 6.1% 1.7% 6.3% 1.4%
nor 0.0% 0.0% 21.4% 0.0%
or 5.7% 0.0% 2.5% 0.0%

Also occurs very infrequently, so the percentages are not very meaningful.
Nor, the negative additive, which occurs with an overall frequency greater
than two only in religion (fourteen times), is used cohesively 21.4% of the
time in that genre (and is never so used in the others). Negative additive con-
junctive cohesion then is an important feature of religious text, and also sug-
gests an important feature of its rhetorical structure: namely, an emphasizing
or focusing gradually in a list, by set reduction. Phrased another way, it might
be said that the high percentage of negative additive conjunctions in this genre
indicates a tendency tow&rafalsification, the most consistent method of proof
(see Popper, 1972). That is, the high frequency of negative additives means
that religious texts proceed through the elimination of alternatives, which is
peculiar, since science is typically the genre which adheres to negative evidence.
But the contradiction of traditional views of science and religion by this find-
ing is not to be disregarded, since traditional views of science have recently
come under flre generally (Foucault, 1972), but science is also seen, subse-
quently in this study, not to rely on other traditionally accepted argumenta-
tion devices, such äs causality.
The last two additives in these texts, and and or, are used structurally (that
is, required by the syntax) much more often than cohesively. For both, how-
ever, it is clear that conjunction initial style is frowned upon in Journalism
and science.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 359

Of the two adversaries, but and yet, the latter occurs cohesively only in
fiction and religion (and again this conjunction does not appear at all in
journalism and science). In this case, however, the cohesive use of this con-
junction is in the majority. Exactly half of its occurrences in fiction are co-
hesive and an even higher 75% of those in religion are cohesive. Its function
is obviously cohesive, but there is also another function operating in these in-
stances, namely emphasis. Making a Statement and then repudiating it through
an adversative can have the effect of emphasizing a. subset characteristic, which
appears to be its principal function here, and which relates to the claim made
above about falsification äs an argumentation device in religion.
The other adversative, but, occurs cohesively in the genres, the following
percentages of the time:

Fiction Journalism Religion Science

Cohesive
but 44.5% 47.9% 24.1% 0.0%

This time, journalism uses an adversative cohesively, whereyet did not oc-
cur at all in that genre. Granted that but can occur structurally (in Halliday
and Hasan's sense), it is clear that its rhetorical function (and it does have a
high overall frequency in all genres but science), is that of contrast and that,
by putting it in sentence initial position (including post-semi-colon), what is
achieved is emphasis through foregrounding. This is the case even with its
high cohesive use, because conjunctions äs a class do not usually appear in
that position.

5.3. Subordinating cohesive conjunction

Not all the subordinating conjunctions listed in Table l behave cohesively.


For example, that, with the exception of the causal conjunction so that, only
behaves structurally, äs a complementizer. Than is cohesive, but not äs a con-
junction: It is cohesive äs a comparative in such constructions äs better than.
Like is a subordinating conjunction in some of the texts, but is very frequent
only in fiction, and it is only in that genre that it is used cohesively (but only
twice out of 45 occurrences). It behaves more like a comparative, however,
than one of Halliday and Hasan's four conjunction categories. Those that do

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
360 RaoulN. Smith and William J. Frawley

behave cohesively äs conjunctions are after, although, äs, because, before, for,
if, since, until, and while. These will be classified into the Halliday and Hasan
semantic typology of additive, adversative, causal, and temporal, and will be
discussed by groups.1

5.3.1 Additive
The only cohesive subordinating conjunction that has an additive sense in
these texts is äs. (It should be noted that it does not appear in Halliday and
Hasan's inventory, however). It is not used cohesively in journalism, but it
does appear äs a cohesive device in the other three genres with the following
similar frequencies:

Fiction Religion Science

Cohesive subordinating conjunction


äs used äs an additive 5.7% 5.7% 5.4%

The use of the additive 'äs' is mainly in phrases such äs 'As can be seen
from,' 'As is the case,' and 'As is well known.' All other occurrences of cohe-
sive äs are in the comparative construction äs... äs.

5.3.2. Adversative
There are two subordinating conjunctions in the texts which are adversative
in sense. These are although and though. Neither is very frequent, and the only
one which behaves cohesively is although. In particular, it is used cohesively
in religion 62.5% of the time (of eight occurrences), and in science 45.5% of
the time (of eleven occurrences).

5.3.3. Causal
In these texts, there are three subordinating conjunctions used cohesively
which have a causal sense. These are because, for, and since. (So, which Halli-
day and Hasan list äs causal, usually occurs in the phrasal conjunction so that,
and occurs alone, cohesively, only once in the entire set of texts, and that is
in the religious sample '... like a satisfying, mouth-watering steak. So it is
spiritually.' Its sense here is not causal but comparative.)
As reported above, religion uses subordinating conjunctive cohesion more
than the other genres. This is especially clear with because and for. These

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 361

have a low frequency in journalism and science and never behave cohesively
in these genres. The other two, however, do have instances of cohesieve use.

Fiction Religion

because 10.0% 31.6%


for 40.0% 77.3%

Since, which could have a temporal sense, but which only occursin a causal
sense, is also used cohesively in these genres, to be precise, 40% of the time in
fiction; 33.3% in religion; and 25% in science. Although not of high frequency
in any of the texts, it has the highest frequency in science (sixteen times), and
is used cohesively in that genre one-fourth of the time. The following table
summarizes the use of causal conjunction in these genres:

Causal subordinating conjunction

Fiction Journalism Religion Science

Cohesive causal 11 0 25 4
Total causal 33 17 47 27
Total subordinating 350 322 509 300

It is clear that causal conjunction is used cohesively most often in religion,


somewhat less in fiction, less in science (and only with since), and rarely in
journalism. We will discuss the rhetorical implications of this below.

5.3.4. Temporal cohesive Subordination


The subordinating conjunctions that are used cohesively and are temporal in
sense in these texts are after, before, until and while. These belong to the
Halliday and Hasan subcategories sequential, preceeding, terminal, and dura-
tive, respectively. Their cohesive use is summarized in the following table.

Fiction Journalism Religion Science

after (sequential) (2/7)28.6% (2/13)15.4% (0/2) 0.0% (1/2)50.0%


before (preceding) (1/8)12.5% (1/4)25.0% (1/5)20.0% (0/5) 0.0%
until (terminal) (0/0) 0.0% (1/3)33.3% (0/4) 0.0% (0/0) 0.0%
while (durative) (0/0) 0.0% (4/10)40.0% (1/12)8.3% (0/0) 0.0%

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
362 RaoulN. Smith and William J. Frawley

As can be seen from this table, none of these occurs with a very high fre-
quency. As a class, temporal conjunctions are less frequent than causals in all
four genres except journalism.

Fiction Journalism Religion Science

Cohesive temporal 3 8 2 1
Total temporal 15 30 23 7
Total subordinating 350 322 509 300

In fiction, there are almost twice äs many causals äs temporals (almost the
reverse of journalism), and in science almost four times äs many.
The preferred use of specific semantic subcategories by various genres is
also interesting. The sequential temporal after occurs most frequently in
journalism, and next most frequently in fiction (rarely in religion or science).
The preceeding temporal, before, is almost evenly distributed among genres.
This seems to imply a preferred focus of time perception, or at least the re-
porting of time-differentiated events. Fiction uses both after and before about
equally, journalism Stresses the sequential temporal rather than the preceding
temporal, and religion and science prefer the reverse order — preceding rather
than sequential. It is also interesting to note that fiction and science do not
use terminal or durative conjunctions in these texts but journalism and reli-
gion do, both t o almost the same extent.
The cohesive use of these temporals, äs with most subordinating conjunc-
tions, is not very high. The average is 17.4% over all four genres, with a ränge
of 8.7% for religion to 26.7% for journalism. So there are genre-specific dif-
ferential usages of temporal conjunctive cohesion. When the two types of
conjunction, causal and temporal, are compared with relationto their cohesive
use, one discovers the following contrasts:

Fiction Journalism Religion Science

Temporal cohesive 20.0% 26.7% 8.7% 14.3%


Causal cohesive 21.4% 6.2% 30.0% 15.4%

Although fiction and science show a similar propensity for the use of these
two conjunction types in a cohesive way, journalism and religion behave the
exact opposite of each other. These facts cannot be accounted for by the total
frequency of each category, however, äs is obvious from the following table:

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 363

Fiction Journalism Religion Science

Total temporal 15 30 23 7
Total causal 28 16 30 26

In all four texts, the additive conjunctions, coordinatingandsubordinating,


are the most frequent (largely because of the frequency ofand), followed by
the adversative and the causals. Causal is much more frequent than temporal
in all genres but Journalism, where the exact opposite holds true. The im-
portance of temporal conjunction over causal in Journalism is a good indica-
tion of the populär Interpretation of Journalisticprose asbeingevent-centered,
at least from a rhetorical point of view. In addition, the low number of causals
may add to the perception of its objectivity.

5.3.5. Hypothetical
If The conjunction // does not appear in the Halliday and Hasan inventory,
nor does it seem to fit any of their categories. Yet is does behave cohesively
in the Corpus äs in:

The snow again. If only the fucken weather wasn't so lousy.

Various senses of//have been reported. The most common meaning found
in the Corpus is that of a hypothetical. Although simplistic,2 this is how it
will be referred to here.

Fiction Journalism Religion Science

Cohesive// 13/50(26.0%) 5/12(23.8%) 12/41(29.3%) 10/21(47.6%)

In terms of absolute occurrences of the hypothetical conjunction, // is


used most frequently in fiction and religion, and to a less, but equal, frequency
in Journalism and science. As a cohesive device, however, it is used most fre-
quently in science. This fact truly forces us to focus our attention on the
function of conjunctions äs cohesive devices. Although these sentence initial
uses of if suggest a theme-rheme or given-new Interpretation, this would be
too simplistic and would account for clause position äs a whole, rather than

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
364 RaoulN. Smith and William J. Frawley

the use of the conjunction. If we return to Halliday and Hasan's discussion of


the concept of conjunctive cohesion, it is clear that (f does function to relate
linguistic elements that are in succession in the text.3 Notice what happens
to the cohesion of the text when the clauses in the preceding sentence are
reversed.

this would be too simplistic and would account for clause position äs a whole
rather than the use of the conjunction. It is clear that if does function to re-
late linguistic elements that are in succession in the text if we return to Halli-
day and Hasan's discussion of the concept of conjunctive cohesion.

The whole second sentence receives a much different Interpretation. The co-
hesive Interpretation is achieved only with the // clause in sentence initial
position, and the reason for this is the cohesive function of the if in that posi-
tion. So if can be used cohesively and since it most commonly has the sense
of a hypothetical, the category of hypothetical should be added to the
Halliday and Hasan inventory.4

5.4. Genre-specific use of conjunctive cohesion

Cohesion is a semantic notion. Its fundamental sense is that of a tie between


two sentences. The tie in the case of conjunctive cohesion is that of ajoining
of two sentences by a conjunction in initial position. But äs discussed above,
specific conjunctions have their own meanings, which add to the semantics of
the conjunctive tie. Our goal in this study (and others currently underway;
for example, Frawley and Smith, 1983) has been to characterize the differ-
ences among various text types in Hnglish. One of the ways öl accomplishing
this, and a specific goal here, is to give a description of the high level concep-
tual structures occurring in these genres. This can be achieved by comparing
from a semantic point of view, the preferred conjunctive forms used in these
texts. We have therefore tabulated the percentage of cohesive uses of con-
junction, both coordinatively and subordinatively, äs a function of their
total occurrences in the respective texts. These percentages, and their rank
ordering by genre, appear in Table 2.
As shown earlier, conjunction äs a text structure-forming device is used
cohesively, on the average, less than 10% of the time. Since we are focusing
on cohesive uses, we are also examining, therefore, only sentential uses and

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 365

Table 2. Cohesive conjunction types (coordinate and subordinate) äs a percentage of


total conjunction. (Rank ordering is indicated in parentheses.)

Fiction Journalism Religion Science

additive 3.93%(2) 0.84%(3) 3.42%(1) 1.48%(1)


adversative 4.51%(1) 3.37%(1) 2.98%(2) 0.67%(3)
causal 1.05%(4) 0.00%(5) 2.18%(3) 0.54%(4)
temporal 0.28%(5) 1.12%(2) 0.00%(5) 0.13%(5)
hypothetical 1.24%(3) 0.70%(4) 1.05%(4) 1.35%(2)

not phrasal ones. In addition, we are examining these cohesive uses äs a pro-
portion of the total use of conjunctions.
The semantic categories of conjunction which we have adopted here,
namely Hailiday and Hasan's additive, adversative, causal, and temporal,
supplemented by our hypothetical, present us with a rieh set of categories for
characterizing the semantic (including the rhetorical) structure of these genres.
These will be considered in order.

5.4.1. Fiction
Cohesive conjunction in fiction can be characterized äs having a high quality
of adversativeness. Characteristics of such text structure in fiction are a high
frequency of but and yet. Their principal function is to emphasize through
contrast. Although and though, which could also function in this way, are
seldom used in fiction.
Fiction is also clearly additive in that and, or, and äs occur quite often,
with approximately the same relative frequency. These t wo, the adversative
and the additive, contribute almost 80% of the cohesive conjunctive text
building in fiction.
The third most important element in fiction texts is that of the hypotheti-
cal — even though comparatively, across genres, this feature in fiction is much
lower than its presence in science, for example. However, given the five
semantic/rhetorical conjunctive devices available to the genre, hypothetical
ranks third.
Causal conjoining is the fourth most important device in fiction and is
characterized by a high frequency of for, since, and because, in order of
descending frequency. The use of this conjunctive device in fiction is second
only to that in religion.
The least employed of the conjunctives in fiction is the temporal, and only

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
366 RaoulN. Smith and William J. Frawley

after (sequential) and before (preceding) occur in this sample. Examples of


the terminal until and the durative while do not occur.
All this suggests the following about fiction äs a discrete genre. It is dis-
tinctly additive and adversative. Its additivity, no doubt, results in a great deal
of right branching, which makes the texts cognitively simpler, and which con-
tributes to the seriation of events intuitively characteristic of the genre.
Adversativity, in fiction, is decidedly not of repudiation, but of foreground-
ing (otherwise adversatives different from 'but' would have appeared). Again,
this seems to be related to the whole notion of the seriation of events, and in
this sense, the adversative 'but' functions much like the additive: fiction, äs
a genre, overwhelmingly coheres through the additive and contrastive linking
of sequential events.
What is not characteristic of this genre are causality, hypotheticality, and
temporality. The absence of causality is understandable from the nature of
narration, however. Causality in narrative is a higher-level plot function and
is manifest through the convergence of the actions of characters or of back-
ground Information (see Grimes, 1975). Thus, causality does not function äs
an explicit cohesive element in narrative, but is an epiphenomenon, a fact
reflected by its absence in our cohesion data.
So, too, is the explanation for the lack of hypotheticality. Fiction by
definition, is hypothetical discourse. Work such äs Searle's (1979) on the
logic and fictive utterances shows that fictive discourse begins with the as-
sumption of hypotheticality: hypotheticality is not something which works
inside fiction to propagate argument or to bring about comparison with 'real'
events. Thus, the use of hypothetical cohesion in fiction is eliminated a priori
since, like causality, it is a higher-level plot function (or even a hyper-plot
function), and is in fiction by default.
The most peculiar feature, however, is the absence of temporal cohesion.
One would think that, in a genre which relies on the temporal sequencing of
events, äs narrative does, fiction would be overtly temporal. However, it
appears that in fiction, temporality is not signalled cohesively, but either
adverbially or verbally. Gutwinski's (1976) study of the fiction of James and
Hemingway bears this out. Fictive discourse appears to effect temporal suc-
cession not by interclausal connection, but by lexical progression (e.g.
watched and saw that .. .') or by adverbial marking of preceding or succeed-
ing events. While it appears that fiction Signals temporality non-cohesively,
it is interesting to note that the only cohesive temporal elements that we have
identified in fiction are before and after, which are generic markers of pre-
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 367

cedence and succession. Any fiction that does not mark through verbal or
adverbial temporals, does so through generic ones.

5.4.2. Journalism
Journalism, like science, but unlike fiction and religion, does not use conjunc-
tions cohesively very often. But it, like fiction, and unlike religion and science,
uses adversatives cohesively more often than additives, and at an even greater
rate than fiction. Its principal realization is the word but (yet does not occur
at all in Journalism) which has its highest cohesive use in this genre.
The second most important cohesive conjunctive in Journalism is the tem-
poral, the conjunctive which ranks the lowest in all the other genres. Half of
the cohesive use of temporals occurs with the durative while, followed by the
sequential after and the preceeding before and the terminal until.
The additive and the hypothetical in Journalism are third and fourth, re-
spectively, in rank order, but are rather close to one another, and neither is
very frequent. Causal conjunction seldom occurs cohesively in Journalism.
The characteristic feature of journalistic texts from the point of view of
conjunctive cohesion can therefore be summarized quite easily. It is character-
ized principally by the adversative use of but, and by the durative and sequen-
tial temporals. It is clear from this, that Journalism places a high value on the
timing of events and, äs compared to historical texts, places emphasis on the
current, durative aspects of these events.
But if we look in some detail at the semantics of these two cohesive devices,
some interesting explanations emerge for their dominance in Journalism. The
unmarked function of 'but' is to indicate an unsatisfied condition (see van Dijk,
1977a), äs opposed to unsatisfied expectation or unexpected consequence,
which are most often manifest äs 'although' or 'yet.' The paucity of the latter
two, and the dominance of but, suggest that Journalism coheres adversatively
through the contrast of conditions in actual worlds, and not by expressing the
violation of expectations or undesired events in possible worlds. That is, the
dominance of the unmarked 'but' is related to the function of Journalism of
describing sequences of events in the world, not the violation of expectations,
which, furthermore, accounts for the lack of hypotheticals in Journalism, since
hypotheticals would entail reasoning about possible worlds. It would be in-
teresting to see what tenses occur in the 'but' clauses: our guess would be a
preponderance of present and past tense, very little perfective aspect to refer
to the time of the utterance, and very little conditional (which are ways by

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
368 RaoulN. Smith and William J. Frawley

which modern English periphrastically achieves the subjunctive or hypo-


thetical).
This insistence on Connectivity in actual worlds, äs exemplified through
'but', is further substantiated by the dominance of temporals. It should be
noted that journalism coheres with a predominance of those connectives which
express the fundamental interval. Before and after bound a temporal interval
in terms of precedence and succession (the end points), and while expresses
the existence of the interval between the end points of before and after. Thus,
these temporal connectives allow journalism the ideal temporal space, an
extended, bounded, temporal interval, which, interestingly, is what journal-
ism is about: the expression of events in time.
Finally, the absence of causality in journalism is also understandable. The
absence of causality, äs said earlier, suggests more objectivity, or at least elim-
inates the suspicion of an imposed order on events. The expression of nec-
essary conditions is not a fact of reportage (which could lead to legal suits,
liability, etc.); rather, the expression of sufficient conditions is a function of
reportage, and this function can be carried by additivity or.adversativity, both
of which are more frequent than strict cause in this genre.

5.4.3. Religion
Religion, like fiction, is rieh in methods used to achieve cohesion through
conjunction. Religion, however, reverses the rank order related to fiction with
respect to additives and adversatives. Religion values the former over the latter
äs a cohesive de vice, and, in particular, the negative additive nor plays a much
larger role than in the other genres.
Adversatives also play a large role in religion. Yet, in particular, occurs
cohesively much more often than in any other genre, and its absolute fre-
quency is much higher. But occurs only half äs often in religion äs it does in
journalism. Although is also used cohesively rather often in religion, to indi-
cate adversativeness.
The role of causals in the cohesive structure of religious texts is also impor-
tant, almost äs important äs the adversatives, and of the four genres, religion
uses causals the most often. The cohesive uses of/or, since, and because are
therefore quite frequent in texts on religion.
Religion and fiction also share a similar proportion of the use of hypothet-
icals äs cohesive conjunctives. Almost one-third of the occurrences of z/, a
rather frequent conjunction in religion, are cohesive.
Of the total number of conjunctions in the religion texts, temporal co-

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 369

hesive conjunctives are minimal. They occur the least in religion and can be
considered rare.
In summary, then, achieving textual cohesion through conjunction is a fre-
quently used tool in religious texts. As to particular conjunctive types, the
negative additive nor, the adversative yety the causals, and hypothetical z/, all
play an important role in structuring religious text. Temporal conjunctives do
not.
From this distribution, two interesting facts about religious discourse
emerge. The first is the distinct connection of religious discourse to fictive
discourse. The two dominant types of cohesion in each genre are adversative
and additive. In another paper (Frawley and Smith, 1983), we have argued
that the pattern of referential cohesion across each of these genres indicates,
or at least provides evidence for, the literary and narrative origins of religious
discourse. Here, again, in conjunctive cohesion, this pattern appears. The strict
seriation of events in each of these discourses is doubtless a critical factor in
their Connectivity.
But religion differs from fiction considerably in the use of causality, tem-
porality, and hypotheticality. The most curious of all of these is the relatively
widespread use of causality in religious discourse. Evidently, the logic of reli-
gious texts demands overt expression of the causal connections across events,
which Stands to reason, given the search for ultimate causes in theology. But
the lack of temporality and the paucity of hypotheticality is peculiar: espe-
cially the paucity of hypotheticality, since religious discourse is fundamentally
philosophical discourse, and one would assume that hypothetical argument
would dominate in this genre. But if it is true that religious discourse derives
from fictive discourse, then this pattern is understandable. Religion and fic-
tion share only the seriation of events. Where religion and fiction differ is in
their respective access to causes and hypotheses. It may, in fact, be true that,
historically, religion separates from fictive discourse by the Implementation
of these very cohesive elements, but this remains a speculation.

5.4.4. Science
Science is like journalism in its avoidance of the use of conjunction in a cohe-
sive manner. As a matter of fact, it is the least-likely genre to use this device
to achieve textual cohesion.
Of the five types of conjunctive cohesion, the additive is the most frequent
in science and it occurs, the few times it does occur, with and.
The hypothetical is the next most frequent cohesive conjunctive in science,
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM
370 RaoulN. Smith and William J. Frawley

the only use of the hypothetical to reach this rank level in all five genres. In
science, almost half of the occurrences of if are cohesive. It clearly plays an
important role in the argument structure of science texts.
The adversative and the causal conjunctives appear cohesively in science
with an almost equal frequency, more than half of that of the hypothetical.
The coordinating but and yet never occur cohesively in science. The only
representative of a cohesive adversative in science is although, almost half of
whose occurrences in that genre are cohesive. As for causals, since is the only
causal conjunction which behaves cohesively. It has the highest of causal fre-
quency in that genre and behaves cohesively there one-fourth of the time. As
a category, science ranks third in its employment ofthat device.
The last of the conjunctives t o be used cohesively in science are the tem-
porals, a class which has an overall low frequency. The only occurrence in the
science texts of seven cohesive uses of this conjunctive type in all the texts is
after, which occurred once.
Science, then, is the least likely genre to use conjunction äs a cohesive de-
vice in text. When it does use it, science is conservative in its use, employing
the additive and the most frequently. The hypothetical // is the next most
frequent conjunctive, followed by the adversative although and causal since.
What does all this mean for science äs a genre? First of all, the predomi-
nance of additivity in science is understandable from Cheong's (1978) argu-
ment, äs made earlier: a genre whose complexity lies in its lexicon compen-
sates for this in recoverability and right branching, äs additive conjunction
entails. Thus, additivity has a low functional load, complementary to the high
functional load of the lexicon in this type of discourse.
Second, the high frequency of hypotheticals is evidence for the use of
'possible worlds' arguments in science. It has, of course, long been known
that science works in some sort of hypothetical manner, whether through
induction, deduction, or abduction. Nonetheless, whatever choice of inferen-
tial strategy, the dominance of hypotheticality shows the necessity in science
of connecting arguments through modal notions; Cheong's (1978) study, in
fact, documents the presence of modals proper in scientiflc texts.
Finally, it is curious to note the distinct paucity of causal connectives in
science. This finding flies in the face of what is traditionally assumed to be
operative in science: the search for ultimate physical causes of phenomena.
There are two explanations for this paucity of causality in scientific texts.
Either science is not the causal discipline it has long been assumed to be, or

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 3 71

science achieves causality in other ways. The evidence points to the latter
subsuming the former.
There are many studies from both a textual and philosophical standpoint
which suggest that science is not causal. Feyerabend (1972), for example,
argues eloquently that science is solipsistic and not universally a discipline
directed toward the discovery of causes, and Cheong (1978) has shown that
the predominant method of causal connection in science is logical, not physi-
cal effect, causality. Thus, given our data and these studies, it appears that
science connects its arguments through a different type of causality, logical,
which makes it appear to be non-causal. The predominance of hypotheticals
is, furthermore, connected to this. Science is a discourse which postulates
causes in possible worlds: the texts reflect this tendency to postulate possible
causes, not actual ones. Thus, science is not such a strictly referential discourse
äs is often assumed, but one which constructs arguments in a hypothetical
manner and favors logical and necessary succession over physically necessary
succession. Here, then, is evidence for the predominance of analytical over
synthetic truth in scientific discourse.

6. Summary and conclusions

It is clear that of the five types of devices available for achieving cohesion in
a text - reference, Substitution, ellipsis, lexical cohesion, and conjunction
(Halliday and Hasan, 1976) - conjunction is not used very often. But it is
also clear from the present study that there are differences in the amount and
type of conjunctive cohesion used by the four genres investigated, and that
the kinds of conjunctive cohesion used in the variousgenre types issignificant.
In addition, it is also clear that fiction and religion are more similar to each
other in the amount and kinds of cohesive conjunction used in generating
these texts, than they are to journalism and science. Why these latter two
genres do not use conjunction cohesively very often may be due to the strong
influence of prescriptive teachings. (Why writers on religion would notslavish-
ly follow the same rubics, however, is not clear.)
The kinds of conjunctive cohesion used in the various genre types is of
extreme importance, however, because its semantics give us an excellent in-
sight into the argument or narrative structure of each text type. This Informa-
tion, when formalized, will hopefully lead to a characterization of genre types

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
372 RaoulN. Smith and William J. Frawley

which will help to drive the text structure and rhetorical components of a text
generating System or a text analysis schenie.

Notes

1. It must be emphasized that the classification of conjunctions äs being coordinating


and subordinating comes from the Brown corpus tags.
2. There are pragmatic features of the use of //based on accepted assumptions on the
part of the Speaker/writer with respect to the hearer/reader (see Caron, 1977).
3. Notice the non-hypothetical use of if in this sentence. and its link to the mood of
the verb in the clause. If the phrase were 4If we were to return . ..,' then it would
be hypothetical (äs is the present sentence).
4. A similar solution, but to a wider set of z/-like conjunctions in publicist prose ap-
pears in Uhlifova (1982: 68). Other classifications of conjunctions have been pro-
posed on a semantic basis. Buchanan (1780), for example, has fourteen categories:
copulative, disjunctive, concessive, adversative, causal, illative or rational, final or
perfective, conditional (including if)t exceptive or respective, diminutive, suspensive
or dubitative, expective, ordinative, and declarative.

References

Buchanan, James (1780). A Regulär English Syntax. Philadelphia: Styner and Cist.
Caron, J. (1977). Essai d'analyse semantique experimentale: la conjonction 4si\ Linguis-
tique et Semiologie, Travaux du Centre de Recherches Linguistiques et Semiologiques,
Universite de Lyon II, pp. 99-126.
Cheong, Lee Kok (1978). The Syntax ofScientific English. Singapore University Press.
Feyerabend, Paul (1972). Against Method: Outline of an Anarchistic Theory of Knowl-
edge. London: New Left.
Foucault, Michel (1972). The Archaeology of Knowledge. New York: Harper and Row.
Frawley, William J. (1980). Discourse analysis of communicative disorders. Language
Sciences 2: 318-338.
Frawley, W. and Smith, R.N. (1983). Patterns of textual cohesion in genre-specific dis-
course. In Linguistics, Humanism, and Computers, Stephanie Williams (ed.). Norwood,
NJ: Ablex.
Givon, Talmy (1979). Understanding Grammar. New York: Academic Press.
Greenbaum, Sidney (1969). Studies in English Adverbial Usage. London: Longman.
Grimes, Joseph (1975). The Thread of Discourse. The Hague: Mouton.
Gutwinski, W. (1976). Cohesion in Literary Texts. The Hague: Mouton.
Halliday M.A.K. and Hasan, R. (1976). Cohesion in English. London: Longman.
Hunt, Kellogg W. (1964). Differences in Grammatical Structures Written at Three Grade
Levels. Florida State University: Tallahassee, FL.
Meyer, Bonnie, J.F. (1975). The Organization of Prose and its Effect on Memory.
Amsterdam: North-Holland.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Conjunctive cohesion in four English genres 373

Petöfi, Janos (1975). Beyond the sentence: Between linguistics and logic. In Style and
Text, H. Ringbom, etal. (eds.). Stockholm: Skriptor.
Popper, Karl (1972). Objective Knowledge. Oxford: Oxford University Press.
Schank, Roger and Abelson, Robert (1977). Scripts, Plans, Goals and Understanding.
Hillsdale,NJ: Erlbaum.
Searle, John (1979). The logical Status of fictionaljliscourse.Jn Contemporary Perspec-
tives in the Philosophy of Language, Peter French, Theodore Ueling, and Howard
Wettstein (eds.), 233-243. Minneapolis: University of Minnesota Press.
Smith, Raoul (ms.). Repetitive collocation in sublanguages.
Uhlifovä, Ludmila (1982). Syntax. In Linguistica II: Kvantitativni characteristiky
soucasne ceske publicistiky, Marie Tesitelovä (ed.), 38-54. Praha: Ceskoslovenskä
akademie ved, Ustav pro jazyk cesky.
Van Dijk, Teun (1977a). Connectives in text grammar and text logic. In Grammars and
Descriptions, T. van Dijk and J. Petöfi (eds.), 11-63. New York: De Gruyter.
- (1977b). Text and Context: Explorations in the Semantics and Pragmatics of Dis-
course. London: Longman.
Wise, Mary Beth and Stewart, Anne (eds.) (1981). Cohesion y enfoque en textos y dis-
cursos. Yarinacocha, Peru: Institute Lingüistico de Verano.

Raoul N. Smith is Senior Member of the Technical Staff at GTE Laboratories, Waltham,
Massachusetts. His current research interests include text structure, artificial intelligence,
computational lexicography, and human factors in computing Systems.

William J. Frawley is Associate Professor and Associate Director of the linguistics Program
at the University of Delaware. Among his research interests are text structure, semantics,
writing, lexicography, and psycholinguistics.

Brought to you by | Universidade Federal do Rio Grande do Norte UFRN


Authenticated
Download Date | 9/16/19 12:44 PM
Brought to you by | Universidade Federal do Rio Grande do Norte UFRN
Authenticated
Download Date | 9/16/19 12:44 PM

Vous aimerez peut-être aussi