Cohesion and coherence

Wolfram Bublitz
University of Augsburg

1. Introduction

Linguists use the two notions of cohesion and coherence to refer to the (linguistically
encoded or just assumed) connectedness of spoken as well as written discourse or text.
Of course, connecting relations also hold among elements of structure within grammat-
ical units such as word, phrase, clause or sentence. But these intra-sentential relations
are different in kind because they are determined by phonological and grammatical
rules and described, inter alia, as syntactic-semantic relations of valency, dependency,
constituency, modification. Cohesion, operating inter-sententially, and coherence are
key notions in text and discourse analysis, as well as in pragmatics because they also
relate to the complex interrelationship between form, meaning and use of linguistic
expressions in specific (social) contexts.
Native speakers have intuitions about which sequences of utterances do or do not
constitute discourse or text. If, by way of an experiment, we deliberately distort a per-
fectly comprehensible and acceptable text by, e.g. changing the order of its utterances
or its linguistic, situational or socio-cultural context, the effect will be one of confusion
on the part of our hearers or readers. They may still understand each individual utter-
ance but not the resulting string of utterances as a whole, i.e. as one unit with a definite
function in its environment. In the eye of the language user who is trying to interpret
them, they do not ‘hang together’ in a reasonable way. I.e., they do not ­display order
and do not form a meaningful gestalt that fits both into the linguistic environment as
well as the social situation, serves the accepted communicative goal and contributes
to the topic at hand; in other words, they are not coherent. Accordingly, the defining
characteristic of such instances of discourse or text is coherence, which itself rests on
text-forming resources such as cohesion and general structural properties determined
by register or genre.
Though both cohesion and coherence refer to meaning resting on relations of
connectedness (between individual propositions and sets of propositions), which may
or may not be linguistically encoded, they are descriptive categories which differ in
kind. Cohesion refers to inter-sentential semantic relations which link current items
with preceding or following ones by lexical and structural means (cf. below). Cohe-
sion is a kind of textual prosody. Since J. R. Firth, who perceived prosodic effects as
phonological colouring, we use prosody to refer to the property of a feature to extend
38 Wolfram Bublitz

its domain, stretch over and affect not just one but several units. Analogously, textual
prosody refers to cohesive colouring involving more than one element in discourse or
text. As cohesion is anchored in its forms, we can argue that it is an invariant, user and
context independent property of a piece of discourse or text. Coherence, on the other
hand, is a cognitive category that depends on the language user’s interpretation and is
not an invariant property of discourse or text.
Though both cohesion and coherence have found their place as key terms in text
and discourse analysis, they still mean different things to different people. Simplifying
matters drastically, we can say that form and structure oriented linguists, who regard a
text as a kind of long sentence, i.e. as a unit beyond the sentence, focus on cohesion as
an essential feature of textuality. Function oriented linguists, on the other hand, who
equate text with any linguistic expression of any length which is used to perform a
specific function, focus on coherence as the defining feature of textuality.

2. Focus on form: Cohesion

The most influential account of cohesion was developed by Halliday and Hasan in
their book Cohesion in English. Their concept of cohesion is “a semantic one” because
it refers “to relations of meaning that exist within the text” (1976: 4) and “enable one
part of the text to function as the context for another” (Halliday & Hasan 1989: 489).
Cohesion manifests itself in linguistic means that appear at the surface level of lan-
guage. An example of such a cohesive means is them in “Wash and core six cooking
apples. Put them into a fireproof dish” (1976: 2), which “presupposes for its interpreta-
tion something other than itself ” (1976: 4), here the expression six cooking apples in
the preceding sentence. “Cohesion”, they argue, “occurs where the interpretation of
some element in the discourse is dependent on that of another. The one presupposes
the other, in the sense that it cannot be effectively decoded except by recourse to it.
When this happens, a relation of cohesion is set up” (1976: 4).
Halliday and Hasan focus on five kinds of cohesive ties between utterances: (1)
reference, (2) substitution, (3) ellipsis, (4) conjunction, and (5) lexical cohesion. (1)
Reference is described as situational or exophoric, resting on means that refer to the
situation at hand (as in Would you like to join me for a cup of tea this afternoon?), and as
textual or endophoric reference, resting on anaphora or cataphora that refer to partici-
pant roles (personal reference as in Doris likes him very much), to (degrees of) proxim-
ity (demonstrative reference as in Joe won’t get tenure. This is what I can’t understand),
or to (degrees of) similarity (comparative reference as in He always eats meat balls with
his fingers. I detest such manners). (2) Substitution involves noun phrases (My beer is
warm. Please get me a chilled one), verb phrases (I have never talked in Jena before. But
 Cohesion and coherence 39

I did in Halle a couple of years ago) and clauses (I trust you know how to open a bottle
of wine? I believe so). (3) Ellipsis is described as “substitution by zero” (1976: 142) and
likewise involves noun phrases, verb phrases and clauses. (4) There are four types of
conjunction: additive (expressed by and, or, furthermore etc.), adversative (but, how-
ever etc.), causal (so, therefore etc.), temporal (then, next, finally etc.). (5) Lexical cohe-
sion, which rests on “identity of reference” between two items, is realized by forms
of reiteration such as word-for-word repetitions, synonyms, superordinates, general
words, so-called collocations, i.e. members of the same lexical field (euro, cent; moun-
tain, climb, peak) and items which, though they are not hyponyms or synonyms in
the language system, are nonetheless used as such in discourse or text, acquiring their
coreferential status only “instantially” (1976: 289).
In the wake of Halliday and Hasan’s proposal, much work has been devoted to
­re-classify and extend their inventory of cohesive means (cf. Martin 1992). Essentially,

all means can be categorized into three types because cohesion stretches three ways:
there is reference to what has been said, to what will be said and also to what could have
been said instead. This three-way distinction corresponds to Roman Jakobson’s well-
known two modes of arrangement of verbal means, selection (paradigmatic choice,
relatedness in the system) and combination (syntagmatic relatedness), to which, for
a higher degree of delicacy, we can add the two vectors of retrospection and prospec-
tion (cf. Hasan 1984). The following overview categorizes cohesive means into classes
according to their prospective or retrospective orientation.
Retrospective cohesive means are proforms (as anaphora), substitutions, syn-
onyms, hyponyms, word-for-word repetitions, general words or labels and second
pair parts of adjacency pairs. Except for the latter, they are based on similarity of form,
structure, content and function and can therefore be labelled ‘forms of repetition’ (or
‘parallelism’). The three essential properties of repetition are quality, quantity and dis-
tribution. Quality is best described as a scale which represents a continuum of fixity
vs. looseness of form and indicates the degree of formal as well as semantic corre-
spondence between the parallel items involved; there is a cline from total equivalence
to paraphrastic substitution. Quantity refers to the length of the repeating item and
is likewise best represented on a scale with a single word (or even phone) at one end
and a string of utterances at the other. Distribution refers to the distance between the
repeated and the repeating item, which can be anywhere between closely adjacent and
considerably removed.
Prospective devices are used to create a slot (to be filled), thereby reducing the
number of options and, generally, setting up expectations. They include proforms (as
cataphora), text structuring discourse markers (incidentally, actually, cf. Lenk 1998)
and gambits (first of all, before I forget, and another thing), and also general nouns or
labels (Francis 1994) like aspect, dilemma, episode, pattern, problem. Labels do not just
40 Wolfram Bublitz

refer to sections but categorize them; in (1) there is both a prospective (stuff) and a
retrospective (things) label:
(1) “I always meant going down to the shops on Saturday to get fresh all the stuff
to get meat and vegetables beefburgers sausages all these things the children
eat for tea”.  (London Lund Corpus 4.3.90 ff)

To these two classes of cohesive means we can add forms of ordered arrangements that
function both prospectively and retrospectively at the same time. They rest on prin-
ciples such as information assessment and iconicity. The former has been described
from a variety of different points of view, prominent among them the Prague School’s
principle of functional sentence perspective, according to which each sentence part is
evaluated for its relative importance for the information carried by the sentence as a
whole and thus for its communicative function within its contextual environment.
Known or given information (i.e. theme or topic status) has a low degree of com-
municative dynamism (because it contributes little to informational progress) while
new information (i.e. rheme or comment status) has a high degree of communicative
dynamism. The assignment of theme and rheme to parts of an utterance has a clear
cohesive function because it sets up links with preceding and following utterances
(cf. Firbas 1992). The second principle of arrangement, iconicity, is easiest illustrated
by Harvey Sacks’ much-quoted example “The baby cried. The mommy picked it up”,
where the temporal sequence of events is reflected in the linear sequence of the two
adjacent utterances. Adjacency based on iconicity is thus another powerful means of
creating cohesive linkage.

3. Cohesion as a condition for coherence

Cohesion should be kept strictly apart from coherence. It is neither a sufficient nor a
necessary condition for coherence. Referring to van Dijk’s example
(2) We will have guests for lunch. Calderón was a great Spanish writer
 (van Dijk 1972: 40)

Edmondson shows that, though the two utterances are not cohesively connected, they
are nonetheless coherent in a context like (3):
(3) Do you know Calderón died exactly 100 years ago today? – Good heavens!
I’d forgotten. The occasion shall not pass unnoticed. We will have guests for
lunch. Calderón was a great Spanish writer. I shall invite Professor Wilson and
Senor Castellano right away. (Edmondson 1981: 13)

Furthermore, in authentic discourse or text, we can encounter long stretches of utter-

ances which, though they are not cohesively connected, are nonetheless accepted as
 Cohesion and coherence 41

coherent texts because they are, e.g., parts of enumerations or cases of dissociated
interior monologue.
However, while cohesion is not a necessary condition of coherence, studies
have shown that discourse and text tend to be cohesive to a greater or lesser extent,
depending on genre. Following a general principle of cooperation, speakers/writers
are anxious to generate cohesion as a means of guiding their recipients’ interpretation
of coherence and thus, ultimately, of securing comprehension. Cohesive means are
cues which ‘signal’ or indicate the preferred line of coherence interpretation. A lack of
cohesive means may disturb the hearer’s/reader’s interpretation of coherence. Easy and
unimpeded interpretation of both discourse and text as coherent depends consider-
ably on the presence of communicative and meta-communicative textual links.
Of course, the equation ‘more cohesion = more coherence’ is not valid per se. And
yet, there are genres, among them everyday face-to-face conversation, in which the
wealth and abundance of cohesive means is quite extraordinary, as in (4):

(4) A well I must admit I feel I mean Edward’s mother and his great and his
grandfather will come up on Christmas Day but I feel somehow the sheer
fact of not having to have to have this really sort of it’s for one thing it does
nark me that it’s so bloody expensive that he won’t eat anything except the
largest most splendid pieces of meat you know
B how annoying
A and it upsets me you know if he needed it I wouldn’t mind
D come to think of it he’s also he’s also an extremely greedy individual who
A yeah so that if you buy enough [f]
D he isn’t satisfied with a normal portion
A for cheese for for three days if he sees it’s there he’ll eat it you know
 (London Lund Corpus 4.3)

Cohesive means come on all levels. In (4) we find on the level of phonetics and pros-
ody alliteration, assonance, rhythm (cf. Couper-Kuhlen 1983) and ‘sound-rows’ or
‘-sequences’ as in cheese – three – he – sees – eat (Sacks 1992). On the lexico-­syntactic
level we notice articles, proforms, deictics, conjunctions, complementizers, labels
(somehow, do, anything, individual) and members of a common script or semantic
field (mother, grandfather; eat, meat, greedy, portion; nark, upset, annoy). And from
a pragmatic and text-analytic viewpoint we recognize speech act pairs (‘statement’ –
‘supportive statement’ accompanied by supportive elements like yes, how annoying,
yeah), hedges, gambits, discourse markers, collocations and tags (well, I mean, sort of,
you know, come to think of it, I must admit, I feel somehow, the sheer fact, I wouldn’t
mind, for one thing). There is an abundance of cohesive means to the point where they
add up to a rather high degree of redundancy. Hearers, however, who have to process
42 Wolfram Bublitz

online, to interpret ongoing talk in actu (without the possibility to look ahead or the
time to recall what was said before in any detail) depend on a certain degree of redun-
dancy. Redundancy helps relieve their memory and gives them time to understand.
In everyday face-to-face discourse, at least, securing comprehension as a principle
ranks higher than avoiding redundancy. Occasionally, speakers may come up with
too many cohesive cues which add up to a degree of redundancy that is no longer
tolerable and can irritate or even annoy the hearer. Such cases of ‘cohesive overkill’
are triggered by the speaker’s misjudgement of the hearer’s interpretive competence
and range of knowledge.
While there is no direct correlation between (lack of) cohesion and (lack of) coher-
ence, to claim that cohesion is not a sufficient condition for coherence is much more
arguable especially if we accept that the speaker’s/writer’s primary motive for using
cohesive means is to help secure coherence. And indeed, the examples which are given
to prove that cohesion alone does not generate coherence, are not very convincing:
(5) The heads of the city’s uniformed services polished their contingency plans for
a strike. Queen Wilhelmina finalized her own plans for the evening. In a nearby
Danish town, two fishmongers exchanged blows. Anders, by far the stronger,
had a cousin in prison. Many criminals are in prison.(Samet & Schank 1984: 63)
The authors claim that (5) is cohesive but not coherent because the utterances
“exhibit […] connections but ‘make no sense at all’” (Samet & Schank 1984: 63).
­However, as we saw when we looked at Edmondson’s reading of the non-cohesive
sequence of utterances in example (2), judgements of coherence, or, for that mat-
ter, incoherence are not invariantly triggered by the text at hand for any hearer or
reader alike. If the text in (5) was authentic, i.e. an instance of naturally occurring
data such as a narrative, its hearers or readers would easily succeed in reaching a
plausible interpretation of coherence by resorting to the larger context, to the situa-
tion at hand, to the overall communicative goal, to their encyclopedic knowledge and
to other sources of supporting data. But even as a piece of constructed, context and
situation free text, (5) allows for a coherent reading. To ask people whether or not an
isolated sequence of utterances ‘has coherence’ is tantamount to asking whether or
not they have enough imagination to come up with a context in which the sequence is
indeed coherent. To reach coherence, they will then quite naturally rely on the cohe-
sive means given in the text. Hence, cohesion is normally a sufficient condition for
coherence because it serves as a powerful and suggestive guideline for the hearer’s/
reader’s interpretation.

4. Focus on meaning: Connectivity

It has been argued that text which is not cohesive can, nonetheless, be coherent ­provided
the propositions underlying its utterances are semantically related to each other. In
 Cohesion and coherence 43

this view, coherence is a semantic notion resting on (a net of) semantic ­relations. The
­variety of semantic or connectivity relations that have been described in the literature
so far can be categorized into a few core classes. Among them we find causality, refer-
ence, coordination (parallel, contrastive), elaboration (example, generalization, para-
phrase), overlap and contiguity (temporal, spatial, aspectual, referential) (cf. Samet &
Schank 1984) as well as so-called scripts, schemata or frames which refer to socially
defined activities and events and help “participants apply their knowledge of the world
to the interpretation of what goes on in an encounter” (Gumperz 1982: 154). Viewing
coherence as a semantic notion usually leads to the assumption that it is a feature of, or
rather in the text. It is ‘there’, ‘in’ the text for people to ‘find’ it. Returning to example (2)
with its interpretation in (3), we can now argue that its coherence rests on the semantic
relations of causality and coordination, which do not only link the propositions of the
two utterances but also two additional though latent, i.e. not realized propositions; here
is a possible paraphrase: We will have guests for lunch – because – we want to celebrate –
because – it is Calderón’s birthday today – and because – Calderón was a great Spanish
writer. The connectivity of (2) is only partly reflected on the surface level of cohesion
by the linear ordering of the two utterances. To make a clear distinction between con-
nectivity and cohesion (with coherence resting on either or on both of these), is justified
on theoretical grounds. In practical discourse analysis, however, the two concepts are
not always easy to keep apart.

5. Semantic connectivity as a condition for coherence

From the fact that coherence is frequently based on semantic connectivity we may
conclude that the latter is both a sufficient and a necessary condition for coherence.
It has been argued, however, that this is too strong a claim. Basically, the argument is
supported by five observations. Firstly, adjacent utterances can be semantically related
without being coherent. In (5) there is connectivity (resting on the semantic relations
of contiguity, time and coordination) but not coherence. Samet & Schank point out
that “our ability to render certain elements of a text understandable by using these
[semantic] connections cannot itself make a text coherent to the point of making
sense” (1984: 64). Secondly, semantic relations can involve utterances which are not
adjacent as, e.g., in enumerations where they are related to a superordinate topic but
not to each other. Thirdly, utterances can be semantically related in ways that are not
clearly identifiable or which allow for alternative identification. Thus, in authentic dis-
course or text it may be difficult or even impossible to figure out exactly which seman-
tic relation is involved, as in (6):

(6) John bought a raincoat. He went shopping yesterday on Queen Street

and it rained.  (Morris & Hirst 1991: 25)
44 Wolfram Bublitz

Each of several connectivity relations is a likely candidate for coherence: “The coher-
ence relation here could be elaboration (on the buying), or explanation (of when, how,
or why), or cause (he bought the raincoat because it was raining out)” (Morris & Hirst
1991: 25). Fourthly, two adjacent utterances can be semantically related (and thus pos-
sibly coherent) for one hearer/reader but not for another, as in (7):
(7) S Well unless you’re not a member, if yer a member of TM people do, ah
simply because it’s such a fucking high price to get in there (1.0 sec) its like
thirty five dollars
C it’s like water polo (2.0 sec)
S Why, is it expensive  (Vuchinich 1977: 246)

This exchange is part of an experiment. In order to test the reaction of his/her inter-
locutor the organiser C makes a deliberately incoherent contribution, i.e. one that is
not semantically related to the preceding utterances. However, as is clear from his/her
reaction, S does see a semantic relation (of comparison) between it’s like water polo and
the preceding text and understands it as coherent (though not straight away, he/she
needs more processing time than usual). Fifthly, two or more utterances can be con-
nected by a semantic relation which, however, cannot be inferred from the linguistic
or the non-linguistic context but only from previously acquired knowledge or from
experience (as is often the case with allusions).
Of course, not one of these five lines of reasoning precludes connectivity from
being at least a major condition for coherence. After all, connectivity rests on seman-
tic relations of the kinds mentioned above rather than on transparence, adjacency or
absoluteness.

6. Coherence: A general view

Coherence is a concept which in its complexity is still not fully understood and a
matter of continuing debate. Though it has doubtless found its place as a key term
in text and discourse analysis, its usage continues to vary to the extent that to give
a comprehensive overview of even the major views advocated goes well beyond the
scope of a handbook article. While after the publication of Halliday & Hasan (1976)
the notion of cohesion was widely welcomed and accepted as a well-defined and
useful category, coherence was often regarded or even occasionally dismissed as a
vague, fuzzy and “rather mystical notion” (Sinclair 1991: 102) with little practical
value for the text or discourse analyst. This view was held by parts of the linguistic
community with, however, some notable exceptions, prominent among them her-
meneutic, ­context- and interpretation-based dynamic concepts of coherence (cf.
below). Since the late ­eighties, there has been renewed interest in the intriguing
 Cohesion and coherence 45

notion of coherence. The remarkable number of almost five hundred titles listed
in a recent bibliography (Bublitz 2010) bears witness to this development and to a
rapidly changing scene in coherence research which is moving away from reducing
coherence to a mere product of (formally represented) cohesion and/or (semantically
established) connectivity.
In her overview, Hellman (1995) distinguishes, inter alia, between approaches
which see coherence (a) “as a formal property of texts” (p. 191ff), (b) “as a discourse
processing concept” (p. 194f), referring primarily to work by Sanders & Spooren(who,
in a 1999 paper, juxtapose the linguists’ view of coherence as a relational concept and
the cognitivists’ view of coherence as a realisation of participant intentions manifest
in each section of a discourse), (c) “as a result of computing referential, causal or
other relations”, i.e. “of a complex problem-solving process in which the reader infers
relations among the ideas, events and states that are described in the text” (p. 195),
and (d) “as a result of computing”, i.e. recognising “the intention(s) of a discourse
producer” (p. 196).

7. A
 hermeneutic, context and interpretation based
view of coherence

Much recent work describes coherence as a mental notion which is interactively nego-
tiated within a given socio-cultural setting and less dependent on the language of
discourse or text itself (cf. the readers by Gernsbacher & Givón 1995; Bublitz et al.
1999). Such a hermeneutic approach describes coherence as a context dependent, user
oriented and comprehension based notion. This view, which dominates work by Fritz
(1982), Brown & Yule (1983) and many others, is in accordance with a fairly long
interpretive tradition in Europe (represented by Ludwig Wittgenstein, Alfred Schütz,
Harold Garfinkel, Hans-Georg Gadamer, Anthony Giddens and others) and the USA.
Within a sociological and ethnographic framework, Dell Hymes, John Gumperz and
others (cf. the reader edited by Bauman & Sherzer 1974) argue for analysing language
within its sociocultural settings. Speakers and hearers alike come to an understand-
ing of the ongoing communicative interaction by linking linguistic and non-linguistic
cues with their background knowledge. They thus continually and jointly negotiate
meaning by constructing a shared context. From such a contextualizing, interpretive
viewpoint, speakers/writers are said to intend, anticipate and (overtly and/or covertly)
suggest coherence while hearers/readers ascribe coherence to utterances within their
linguistic, situational and socio-cultural context.
Because much recent research into coherence follows a context-, negotiaton-,
interpretation-dependent view of coherence, a snapshot account of it seems indicated.
According to this view, coherence is not a discourse or text inherent property, i.e. it is
46 Wolfram Bublitz

not given in discourse or text independently of interpretation. Consequently, one can-

not say “a text has coherence” in the same way as one can say “a text has a beginning
or an end”, or indeed, “a text has cohesion” (the latter being a text inherent property).
We can only say “someone understands a text as coherent”. Of course, coherence is
based on the language of the text in the same way as it is based on other information
provided by the linguistic context, the socio-cultural environment, the valid commu-
nicative principles and maxims, the interpreter’s encyclopedic knowledge etc.
Since it is not texts that cohere but rather people who make texts cohere, we can
say that for one and the same text there exist a speaker’s/writer’s, a hearer’s/reader’s and
an analyst’s coherence, which may or may not match. Typically, different interpreta-
tions of the coherence of a text depend on its linguistic complexity, the temporal, local
and social setting, the interpreter’s familiarity with genre and content as well as his/
her knowledge of the speaker’s/writer’s background (motives, preferences, interests).
Being dependent on interpretation means that speakers/writers can never produce
coherence which is binding for hearers/readers. It is the latter who have to arrive at
their own understanding of coherence. Normally speakers/writers are set to help cre-
ate coherence by (more or less subtly) guiding their hearers/readers to a suggested line
of understanding. Conversely, hearers/readers use these guiding signals as instruc-
tions to align their interpretations with what they take to be the speakers’/writers’
intentions. Hearers/readers assemble and subsequently test a view of coherence which
they assume comes closest to that of the speaker/writer. Hence, coherence is rarely
static but frequently dynamic, i.e. a process rather than a state. It can be tentative and
temporary because it is continually checked against any new information which may
make adaption and updating necessary. Of course, eventually, coherence (especially of
written texts) can lose some of its provisional and temporary character and acquire a
higher degree of permanence.
Coherence is the outcome of the language user’s gestalt creating power. People
are driven by a strong desire to identify forms, relations, connections which they
can maximize in order to turn fragments into whole gestalts, i.e. to ‘see’ coherence in
strings of utterances. Coherence is also a cooperative achievement (in ongoing dis-
course more than in ‘petrified’ text) because it depends on both the speaker’s/writer’s
and the hearer’s/reader’s willingness to negotiate coherence. Mutual understanding
not only rests on the participants sharing the same socio-cultural background, the
same range of knowledge and communicative assumptions, but also on their ability
to figure out unshared experience, i.e. to adjust their own world-view to that of their
interlocutors’. Hearers/readers are constantly engaged in trying to re-create coherence
as an equivalent of the speaker’s/writer’s coherence, but despite their efforts they can
never succeed in coming up with an exact copy. Coherence is only approximate and a
matter of degree and best described as a scalar notion. Any interpretation of coherence
is restricted and, accordingly, partial to different degrees.
 Cohesion and coherence 47

8. Coherence as a default assumption

As a rule, speakers/writers and hearers/readers alike operate on a standard or default

assumption of coherence. This explains why despite the fact that ascribing coherence
frequently involves making difficult and complex acts of inference, cases of disturbed
coherence are not abundant, and why people try to understand coherence even though
the data to go by may look sadly insufficient. As long as there is no evidence to the
contrary, they proceed from the assumption that what they hear and read can be made
coherent, even if this involves making extremely remote and unlikely connections.
Speakers/writers likewise rely on the default principle and expect hearers/readers to
assume that what they say or write is coherent. The default principle of coherence is
accepted for two reasons. Firstly, it relates to a more general strategy according to
which it is normally assumed by the interactants that hearers/readers “do as little pro-
cessing as possible” (Brown & Yule 1983: 60), opting for the obvious and likely reading.
Secondly, it follows from the general principle of cooperation (in the Gricean sense)
and is therefore an essential normative basis of communication, which leads to ratio-
nal behaviour as the only path to efficient communication. (Cf. also Linde’s observa-
tion that “the process of creating coherence” is “a social obligation” 1993: 16).

9. Perspectives

This overview can only touch upon a limited number of methodological and theoreti-
cal approaches to the description of cohesion and coherence, on the one hand, and of
ways and means of (re-)creating them, on the other. For some time now there have been
several common tendencies, among them the tendency to describe a far larger inven-
tory of cohesive means than originally proposed by Halliday and Hasan, the tendency
to refrain from accepting only one canonic definition of coherence, and the tendency
to observe a basic stock of fundamental theoretical and methodological assumptions.
Much recent work demands such descriptive principles as the need to rely on authentic
data, to proceed in an interdisciplinary way, to assume a cultural and “common sense”
basis for coherence (cf. Linde 1993: 192ff), to relate micro-linguistic to macro-linguistic
(e.g. socio-cultural) issues, and to focus on the powerful coherence-securing role of
such (sometimes long neglected) means as gestures (cf. McNeill & Levy 1993), dis-
course topics, collocational orientation.
Discourse topic and the various procedures of handling it (introducing, changing,
shifting, digressing from it) are prevalent and strong means of (re-)creating coherence
(cf. Geluykens 1999). In many societies, speakers/writers are expected to stick to a
topic. Taking up social space is only justified if an utterance is evidently a contribution
to the topic at hand, i.e. an acceptable answer to “the generic question […] ‘why that
48 Wolfram Bublitz

now’” (Schegloff 1990: 55). The relationship between discourse topic and overarch-
ing communicative goals is one of the stubborn problems that future research has to
address. Collocations based on shared polarity, following from the node’s negative or
positive semantic prosody, contribute to coherence because they are likely to recur
in the local as well as global environment. The semantic orientation of the individual
collocation is then not regarded as a single occurrence with no bearing on context but
rather as indicating the overall tenor of the discourse, i.e. its general semantic orienta-
tion. On account of such predictive force, they are regularly instrumental in dispersing
meaning in discourse (cf. Bublitz 1996).
Future research will also have to deal with unsolved problems such as how means
of securing coherence (and ways of describing them) vary from spoken to written
language, from genre to genre, from text type to text type (cf., e.g., Fritz 1999 on
coherence in hypertext), from one society to another, from earlier to later stages of a
language (the diachronic perspective), and from earlier to later stages of acquiring a
language (both as a native speaker and a foreign language learner).

References

Bauman, R. & J. Sherzer (eds.) (1974). Explorations in the Ethnography of Speaking. Cambridge
­University Press.
Brown, G. & G. Yule (1983). Discourse analysis. Cambridge University Press.
Bublitz, W. (1996). Semantic prosody and cohesive company. Leuvense Bijdragen 85: 1–32.
Bublitz, W. (2010). A bibliography of coherence and cohesion. http://www.philhist.uniaugsburg.de/de/
lehrstuehle/anglistik/sprachwissenschaft/downloads_intern/Bibliography_of_Coherence.pdf
Bublitz, W., U. Lenk & E. Ventola (eds.) (1999). Coherence in spoken and written discourse: how to
create it and how to describe it. John Benjamins.
Couper-Kuhlen, E. (1983). Intonatorische Kohäsion. Zeitschrift für Literaturwissenschaft und Lin-
guistik 49: 74–100.
Edmondson, W. (1981). Spoken discourse. Longman.
Francis, G. (1994). Labelling discourse: an aspect of nominal-group lexical cohesion. In M. Coulthard
(ed.) Advances in written text analysis: 83–101. Routledge.
Firbas, J. (1992). Functional sentence perspective in written and spoken communication. Cambridge
University Press.
Fritz, G. (1982). Kohärenz. Narr.
——— (1999). Coherence in hypertext. In W. Bublitz et al. (eds.): 221–232.
Geluykens, R. (1999). It takes two to cohere: the collaborative dimension of topical coherence in
conversation. In W. Bublitz et al. (eds.): 33–53.
Gernsbacher, M.A. & T. Givón (eds.) (1995). Coherence in spontaneous text. John Benjamins.
Gumperz, J.J. (1982). Discourse strategies. Cambridge University Press.
Halliday, M.A.K. & R. Hasan (1976). Cohesion in English. Longman.
——— (1989). Language, context, and text: aspects of language in a social-semiotic perspective. Oxford
University Press.
 Cohesion and coherence 49

Hasan, R. (1984). Coherence and cohesive harmony. In J. Flood (ed.) Understanding reading
­comprehension: 181–219. International Reading Association.
Hellman, C. (1995). The notion of coherence in discourse. In G. Rickheit & C. Habel (eds.) Focus and
coherence in discourse processing: 190–202. de Gruyter.
Lenk, U. (1998). Marking discourse coherence. Narr.
Linde, C. (1993). Life stories. The creation of coherence. Oxford University Press.
Martin, J.R. (1992). English text. System and structure. John Benjamins.
Mcneill, D. & E. T. Levy (1993). Cohesion and gesture. Discourse Processes 16: 363–386.
Morris, J. & G. Hirst (1991). Lexical cohesion computed by thesaural relations as an indicator of the
structure of text. Computational Linguistics 17: 21–48.
Sacks, H. (1992). Lectures on conversation. Blackwell.
Samet, J. & R. Schank (1984). Coherence and connectivity. Linguistics and Philosophy 7: 57–82.
Sanders, T. & W. Spooren (1999). Communicative intentions and coherence relations. In W. Bublitz
et al. (eds.): 235–250.
Schegloff, E. (1990). On the organization of sequences as a source of ‘coherence’ in talk-in-interac-
tion. In B. Dorval (ed.) Conversational organization and its development: 51–78. Ablex.
Sinclair, J. (1991). Corpus, concordance, collocation. Oxford University Press.
Van Dijk, T. (1972). Some aspects of text grammars. Mouton.
Vuchinich, S. (1977). Elements of cohesion between turns in ordinary conversation. Semiotics 20:
229–257.