Beruflich Dokumente
Kultur Dokumente
Associate Editors:
Jacob L. Mey
(Odense University)
Herman Parret
(Belgian National Science Foundation, Universities of Louvain and Antwerp)
Jef Verschueren
(Belgian National Science Foundation, University of Antwerp)
Editorial Address:
Justus Liebig University Giessen, English Department
Otto-Behaghel-Strasse 10, D-35394 Giessen, Germany
e-mail: andreas.jucker@anglistik.uni-giessen.de
Editorial Board:
Shoshana Blum-Kulka {Hebrew University of Jerusalem)
Chris Butler (University College of Ripon and York)
Jean Caron {Université de Poitiers); Robyn Carston (University College London)
Bruce Fraser (Boston University); John Heritage (University of California at Los Angeles)
David Holdcroft {University of Leeds); Sachiko Ide {Japan Women's University)
Catherine Kerbrat-Orecchioni (University of Lyon 2)
Claudia de Lemos (University of Campinas, Brasil); Marina Sbisà {University of Trieste)
Emanuel Schegloff (University of California at Los Angeles)
Paul O. Takahara (Kobe City University of Foreign Studies)
Sandra Thompson (University of California at Santa Barbara)
Teun A. Van Dijk (University of Amsterdam); Richard Watts (University of Bern)
63
Wolfram Bublitz, Uta Lenk and Eija Ventola (eds)
Coherence in Spoken and Written Discourse.
How to create it and how to describe it
COHERENCE IN
SPOKEN AND
WRITTEN DISCOURSE
HOW TO CREATE IT AND
HOW TO DESCRIBE IT
SELECTED PAPERS FROM THE
INTERNATIONAL WORKSHOP ON COHERENCE,
AUGSBURG, 24-27 APRIL 1997
WOLFRAM BUBLITZ
Universität Augsburg
UTA LENK
Universität Augsburg
EIJA VENTOLA
Martin-Luther Universität
Index 297
Acknowledgements
A volume such as this stands or falls by the papers it presents. We would
therefore like to thank the authors of the articles here collected, not only for the
hard work they put in for the completion of this volume, but also for their lively
and interesting contributions to the discussions during the workshop. We also
wish to thank those colleagues who participated in the workshop but whose
papers for various reasons do not appear in this reader: Andrea Gerbig, Gisela
Redeker, Sorin Stati, Edda Weigand, and the late Maria-Elisabeth Conte
(University of Pavia), whose premature death a few months after our workshop
prevented her from writing up her contribution on Semantic integration devices
in texts, which had received much acclaim and positive comments at
presentation.
Our special thanks go to Carla Bazzanella, who was involved in developing the
first plan for this workshop, and to Gudrun Nelle, Sarah Gietl and Nils Engel,
the always reliable and eager groundcrew who were the efficient organizers
behind the scenes. Without them, the workshop would not have materialized.
In editing this reader we have received valuable help from Carol LeRoux, who
checked the non-native speakers' English, Gudrun Nelle, a reliable proof-reader,
Sarah Gietl, our desk-top publisher-in-chief, and, in the final stages, Thomas
Henrichs, our computer wizard.
Last, but not least, we want to thank the Gesellschaft der Freunde der
Universität Augsburg and the Deutsche Forschungsgemeinschaft for their
financial support which made the workshop possible, a gathering which was
experienced by all who came as most delightful and interesting.
About the authors
Carla Bazzanella is Professor of the Philosophy of Language at the University
of Torino, Italy. She has published on pragmatics, morphosyntax, socio-
linguistics, and applied linguistics. Her most recent monograph is Le Facce del
Parlare (1994) and she has edited Repetition in Dialogue (1996).
dependent on the language of the text itself. This view of coherence underlies
an increasing number of recent analyses of authentic data which relate micro-
linguistic to macro-linguistic issues and in which cohesion often plays merely a
minor role.
The tendency to observe a basic stock of fundamental descriptive
assumptions while refraining from accepting one canonic definition of
coherence is also reflected in this reader, which consists solely of papers
presented at the 'Workshop on Coherence in Discourse', held at the University
of Augsburg in April 1997. In the invitation to the workshop, we spelt out our
essentially hermeneutic understanding of coherence, which is outlined in the
following snapshot account.
In accordance with a fairly long tradition on the European continent,
though not in Anglo-American work, we use coherence as a context-dependent,
hearer- (or reader-) oriented and comprehension-based, interpretive notion.
Coherence is intrinsically indeterminate because it is relative to the way in
which language users ascribe their understanding to what-they-hear (or what-
they-read). Accordingly, coherence is not a text-inherent property at all (as are
cohesion and connectivity). It is not given in the text invariantly and
independently of an interpretation, but rather 'comes out' of the text in the
sense that it is based on the language of the text, in the same way as it is based
on additional information provided inter alia by the linguistic context, the
socio-cultural environment, the valid communicative principles and maxims
and the interpreter's encyclopedic knowledge. Since it is not texts but rather
people that cohere when understanding texts (be it at the producing, receiving
or, indeed, analysing side), it can be said that for one and the same text there
exist a speaker's (or writer's), a hearer's (or reader's) and an analyst's
coherence, which may or may not match. However, normally speakers (or
writers) are set to help create coherence by (more or less subtly) guiding their
hearers (or readers) to a suggested line of understanding which comes close to
or, ideally, even matches their own. Conversely, hearers (or readers) use these
guiding signals as instructions to re-create coherence and to re-align their
interpretation with what they take to be the speakers' (or writers') intentions.
To put it differently, they assemble and subsequently test a view of coherence
which they assume comes closest to that of the speaker (or writer). Hence,
coherence is not a state but a process, helped along by a host of interacting
factors situated on all levels of communication (from prosodic variation to
textual organisation, from topic progression to knowledge alignment). As a
process, coherence is not taken for granted but, depending on situation, genre
or text type, rather viewed as being more or less tentative and temporary,
continually in need of being checked against new information which may make
adaption and updating necessary. Eventually, coherence can, especially in
Introduction: Views of Coherence 3
retrospect, lose some of its processural and temporary character and acquire a
higher degree of permanence. Some of the following papers address a further
aspect of coherence: that it is also a cooperative achievement because it
depends on both the speaker's (or writer's) and the hearer's (or reader's)
willingness to negotiate coherence (in the same way as they negotiate
ideational meaning and illocutionary force). 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 interlocutor's. Adopting Reddy's (1979) toolmakers metaphor we can say
that hearers (or readers) are constantly engaged in trying to re-create coherence
as a 'replica' of the speaker's (or writer's) coherence, but despite their efforts
they can never succeed in coming up with an exact replica. Thus, coherence is
only approximate and a matter of degree and best described as a scalar notion.
Any interpretation of the coherence of a text is restricted and, accordingly,
partial in different degrees. Nevertheless, partial coherence rarely turns into
incoherence because, as a rule, participants operate on a generally shared
default assumption of coherence (cf. Bublitz and Lenk, this volume).
To somewhat differing degrees, the concept of coherence proposed
above is shared by the contributors to this collection. Furthermore, lying
behind most essays are several other shared assumptions: that studies on
coherence must be based on authentic, non-fabricated data (using corpora of
spoken or written discourse); that when describing coherence, it is imperative
to sharply distinguish between the producer's, the recipient's and the analysing
observer's point of view; that coherence is medium- and genre-specific, i.e.
that the strategies and means used by speakers or writers to suggest coherence
can (and often do) vary from spoken to written language, from genre to genre,
from text-type to text-type. All papers are original and unpublished works
reflecting contemporary, state of the art trends in coherence research. They
critically re-examine coherence from a variety of different angles, propose new
ways of thinking about it and point out methodological and descriptive
desiderata and caveats that an adequate theory of coherence should eventually
account for.
Included in this volume are 13 of the 18 papers read at the workshop by
contributors whose work over the years has proved to be a major stimulus for
coherence research, among them many of the leading figures in discourse
analysis and pragmatics. The contributions have been deliberately organised
into sections in order to create topical coherence for the reader.
The six papers of the first section {How to (re-)create coherence: means
of coherence) contribute answers to the question of how exactly various
linguistic strategies and means (among them inter alia prosodic variation, topic
4 Wolfram Bublitz
Goffman has pointed out that interlocutors in the course of any natural
conversation are constantly changing the footing of their talk. In Goffman's
usage, this term refers to the alignment which speakers take up to themselves
and to others as evidenced by the way they handle the production and reception
of utterances (1981: 128). Changes in footing may involve different reception
roles or different production roles or both (Goffman 1981: 226ff; also Levinson
1988), and they are commonly understood to be signalled inter alia by prosodic
cues and code-switching, which contextualise the particular footing or
participant framework currently relevant (Gumperz 1982; Tannen (ed) 1993).
Yet precisely how this contextualisation is accomplished and what specific
contribution prosody makes to the 'management' of footing has not yet been
fully spelled out1 - at least not for all types of shift.
The present paper addresses one of the most frequent shifts of footing,
namely that occasioned by the use of reported speech in conversation. What
happens with reported speech is that the unity within a single speaker of the
three production roles which Goffman identifies - animator, author and principal
- dissolves, leaving the role of animator separate from, and independent of,
those of author and/or principal (cf. also Seidlhofer and Widdowson, this
volume). The 'reporting' speaker animates or voices a 'reported' figure without
necessarily composing the words which this figure is made to utter or espousing
the beliefs which the figure's words will be heard as attesting to. 2 The question
which the 'voicing' of figures raises for a prosodist is whether and to what
extent the speaker's phonatory voice is instrumental in the process.3 Using a
methodology developed by crossing prosodic analysis with conversation
analysis (Couper-Kuhlen and Selting 1996), this paper attempts to pin down
12 Elizabeth Couper-Kuhlen
(1)
Toxic chocolate (38/22.16)
Ann and a girlfriend are spending the weekend as guests at Joy's house. The
following exchange takes place over lunch.
(The small f signals that a reportedfigureis being voiced somewhere within the
line, different indices representing different figures. The hyphen signals a return to
the reporting situation.)
1 J: oh and on Sunday,
you can open the first door of the advent calendar.
All: aahh oohh
J: «p>\ never had one before.>
5 A: did you buy it?
J: mhm,
A: you didn't buy the kind with chocolates in it?
J: no.
All: huh huh huh huh
10 J: f1 I didn't think I needed any « / > chocolate (thing).>
fl it's a « / > to:xin you know.>
A: - who're you talking to (.)
talking about.
J: there-
15 f1 « / > chocolate is toxic.>
-> A: - yeah but you said that like somebody says that.
J: no that's the (.) cancer pre
cancer pre- [ventative type.
A: [aaah.
20 J: I honestly cannot fathom; (.)
following that diet;
just to prevent cancer.
Joy presents opening the first door of the advent calendar as a special
treat for her guests (lines 1-2). In fact, even the calendar itself takes on special
14 Elizabeth Couper-Kuhlen
status when Joy claims that it is her first (line 4). Ann expands on this topic by
asking whether Joy bought the calendar. (Buying an advent calendar is to be
understood as contrasted with making one, a custom common in Germany,
where the conversation takes place.) When Joy acknowledges that she did buy
it, Ann expands the sequence again by asking you didn't buy the kind with
chocolates in it (line 7), subtly implying that this kind of calendar might be more
desirable. Thus when Joy now states I didn't think I needed any chocolate
(thing) and it' 's a toxin you know, she is heard as justifying her purchase, an
activity prompted by Ann's treatment of buying a calendar without chocolates
as remarkable and therefore 'accountable'.
Joy's accounts are couched as a warning implying that chocolate is a
health risk (chocolate (thing). it's a toxin you know, lines 10-11). This warning,
however, is presented in someone else's voice, as Ann herself observes: you
said that like somebody says that (line 16). Yet Joy's 'voicing' in lines l0ff
manifestly creates a problem for Ann, because she initiates repair immediately
thereafter: who're you talking to () talking about (lines 12-13). Note that it is
not the content of the utterance which is at issue here. It is true that Joy initially
treats the problem for which repair is initiated as referential in nature. She
responds with a partial repeat, making the reference of it explicit: chocolate is
toxic (line 15). However, Ann's yeah but in next turn (line 16) shows that what
was said is not the point. The issue is the way the utterance was said. As Ann
puts it, the utterance 'sounded like' someone else. What is her interpretation
based on?
Deconstructing Ann's observation leads to a consideration of the nature
of vocal deixis. Prosodic and paralinguistic effects are in fact deictic to a certain
extent: they involve speaking within a given range of relative loudness, pitch
and tempo (Laver 1994) and with a given voice quality (Laver 1990). In the
default case, loudness, pitch and tempo ranges together with voice quality are
anchored to the prosodic/paralinguistic habitus of the speaker. That is, speakers
are accustomed to deploy, and their interlocutors are accustomed to expect,
certain prosodic and paralinguistic 'reference values'. A noticeable shift of these
values - using a pitch, loudness or tempo range or a voice quality which departs
from the speaker's habitus - will be heard as shifted deixis and can evoke the
presence of a second deictic centre. It is precisely this kind of shift which is
hearable in Joy's chocolate (thing) (line 10), toxin you know (line 11) and
chocolate is toxic (line 15): in each phrase she drops into low pitch register
shading off into a final 'vocal fry' or glottal creak. Fragment (1) thus provides
demonstrable evidence of the fact that a figure can be 'voiced' by the way in
which an utterance is configured prosodically and paralinguistically. At the same
time it suggests that the question underlying coherence must be expanded to
why that now and in that way?
Coherent Voicing 15
(2)
Political contradiction
Bill is telling his girlfriend Gina about a 'real political ' discussion which he had
on the beach with someone he had just met
1 B: You know and it's so funny cause he's-
he's a Catholic?
()
and::
you know
like I nailed him on the contradiction;
he's like pro-capital punishment,
but- and pro life,
hhh
10 G: uh huh,
B: fl I said try to explain that to me.
fl I don't underst(h)a(h)nd(h).
fl heh heh
G: - (wait) you said he's:
15 pro:: capital punishment -
B: and pro life,
G: and pro life.
ri:::ght.
B: that's like
20 to me that's like a hu::ge contradiction?
(.)
and he said
?f2 wellhe
he- justified it as (.)
see
25 ?f2 th- those little inf-
?f2 the little (.) uhm
?f2 embryos?
0
?f2 they don't have a (.) decis-
30 ?f2 they can't make a decision.
?f2 you know
16 Elizabeth Couper-Kuhlen
It may also occur if the answer to the question Whose 'other voice ' is this? is
unclear, as we remember from example (1).
Joy's it's a toxin you know (line 11) is followed in the next turn by a
repair initiator from Ann: who're you talking to (line 12). Following a short
pause, talking to is self-repaired to talking about (line 13). Ann's problem thus
is now displayed as being that she does not know who Joy is impersonating
(although she does know that Joy is impersonating someone). It is only once
Joy has specified what figure she has in mind that's the cancer pre- cancer
preventative type (lines 17-18) that Ann acknowledges the repair, her aaah in
line 21 attesting to a changed state of knowledge (Heritage 1984).
A third type of 'trouble' occurs in reported speech sequences when the
answer to the question How is this 'other voice ' being done?, or What is the
speaker doing with this 'other voice '? , is unclear. Figures are always animated
for a particular purpose in situated interaction. It is often the way the voices are
formatted prosodically and paralinguistically which contextualises what they are
doing, or rather what current speaker is doing with them. Where this is unclear,
participants may find it necessary to initiate repair. Please, consider example (2)
once again, where, in re-enacting the exchange he has had with his new
acquaintance, Bill first introduces himself as a figure in the story and animates
this 'voice': try to explain that to me. I don't underst(h)a(h)nd(h) heh heh (lines
12-14). The animation is noticeable first by shifts in personal and temporal
deixis: Bill's interlocutor becomes you, to whom the request try to explain that
to me is directed, and the speaker's account for this request I don't understand
is anchored to the moment of speaking in the reported situation. But there is
also a sign of shifted vocal deixis in the breathiness superimposed on the
figure's account I don't underst(h)a(h)nd(h) (line 12).8 The laughter particles in
line 13 are compatible with such a shift.
The parahnguistic formatting of Bill-the-figure's voice is interpretable as
Bill-the-narrator 'doing' something. But Gina manifestly has a problem in
determining what exactly he is doing with Bill-the-figure's voice. Her problem
becomes clear when she initiates repair in next turn: (wait) you said he's: (line
14). Her initiation is accompanied by a candidate repair pro:: capital
punishment (line 15), which Bill acknowledges by completing with and pro life
(line 16). Gina ratifies this completion and pro life (line 17), signalling with a
follow-up ri:::ght (line 18) that she has got the point. Moreover, Bill now
makes explicit in the next turn what his point is: that's like - to me that's like a
hu::ge contradiction (lines 19-20). It is thus an understanding of the
contradiction between the two political positions referred to which is treated as
being necessary in order for Gina to make sense of Bill-the-figure's turn. Once it
is clear that there is a contradiction and what it is, the parahnguistic overlay of
breathiness and laughter fits in as a contextualisation cue to how Bill-the-figure
18 Elizabeth Couper-Kuhlen
evaluates the contradiction (as well as the person who espouses it) and to how
Bill-the-narrator wishes his addressee to evaluate it.
To summarise the argument so far: by examining reported speech
sequences in which there is 'trouble', as evidenced by participants finding it
necessary to initiate repair, we have shown that the pursuit of coherence in
conversational reported speech involves finding plausible answers to at least
three questions: Is this current speaker's 'voice ' or someone else's?; If so,
whose 'other voice' is this? and How is this 'other voice' being done? (or
What is the speaker doing with this 'other voice '?). 'Trouble' may occur when
there are no clear answers or only conflicting answers to these questions.
Moreover, finding appropriate answers sometimes depends crucially on
prosodic and paralinguistic framing devices. Where prosodic and paralinguistic
signals are inappropriate or ambiguous (and verbal content does not fully
disambiguate), participants may find that repair is required in order to establish
coherence in reported speech sequences.
The question which now arises is what counts as appropriate prosodie and
paralinguistic framing for conversational reported speech. To address this
question we shall examine a selection of'successful' reported speech sequences
and compare them to the repairable ones above.
(3)
Rainbows (12/739)
Two high school friends Janet and Ann, reunited after several years, are recalling
their college experiences. Janet's husband Steve is present
Janet brings two figures on stage here, the 'professor from Carleton' (line
1) and a 'guy in class' (line 4). In line 12, with no further quotative introduction,
she animates the figure of the student. Moreover, her interlocutor Ann orients
to the presence of this 'other voice' with her high WHAT? in line 15.9 Although
the token what functions as a next-turn repair initiator with respect to prior talk
in some contexts, Janet does not treat it this way here. She does not respond by
recycling her turn but carries on instead with (because) God made the rainbow:
In fact, Ann's what token is cued with loud volume and high pitch - a display of
astonishment (Selting 1996), in this case strongly projective of disagreement.
Such strong disagreement would hardly be appropriate, were Janet's talk to be
understood as being 'in her own voice'. But it is fully appropriate if there were
no rainbows before the flood is animated as belonging to someone else.
How is it that Ann recognises lines 12ff as being 'in another voice'? In
this case there is no significant change in the voice quality, pitch or loudness of
Janet's speech. What does distinguish these lines from surrounding talk,
however, is their pronounced rhythmic quality. The accented syllables are timed
Coherent Voicing 21
(4)
Galileo (12/370)
Same speaker constellation as in (3). Talk here is about the Catholic church and
the fact that its doctrines change very slowly
her accent on probably (line 33) hits exactly the pitch level that Janet's prior
animated onsets had and the next accent on ro- is targeted to step down from
there. The pitch replication now breaks down with what is manifestly a slip of
the tongue: ro(und) instead of world (cf. line 33) but the rhythmic replication
persists with the accumulation of accents in the world is round (line 34). With
her prosodic 'chiming in', Ann thus implicitly signals her understanding of the
presence of some 'other voice' and indeed of some particular 'other voice'.
Moreover, she can also be heard to co-align with the 'take' which current
speaker has on this voice (see below).
When more than one figure is 'on stage' in a reported speech sequence, the task
which conversationalists face is all the more complex. If the figures' voices are
prosodically and paralinguistically animated, they must be done in such a way
that they are not only recognisably distinct from the reporter's own voice but
also distinct from one another. Reconsider fragment (3), which demonstrates
one way conversationalists have of accomplishing this task.
Following her animation of the student in lines 12, 14 and 16, Janet begins
to animate the professor in line 19 with I 'm sorry. Significantly, the fact that this
line is the professor's voice and not the student's is inferable only from its
distinctly different prosodic delivery. (The anaphoric expression in Janet's
quotative and he's like (line 18) could refer to either the student or the
professor.) Instead of the rhythmic and loud configuration which characterises
the student's there were no rainbows (line 12) and God made the rainbow (line
17), line 19 - I'm sorry - has lower volume, stylised pitch and syllable
stretching. It is because of this markedly different prosody that we infer that it is
not the student who is 'apologising' but some other figure.
Just as with the prosodie framing of a single figure, the voicing of two or
more figures also displays individualised recipient design. That is, different
figures receive not only distinct voicing but also specific different voicing.
Consider now another case in which this is successfully accomplished:
(5)
Rollerblades (12a/40)
Bret and Wanda are brother and sister. Ricky is Wanda's young son. The topic
here is rollerb lading.
1 B: well I mean
Here you see
you know five-year-old kids doing -
24 Elizabeth Couper-Kuhlen
W: well Ricky
5 some of Ricky's friends
and they do wonderful -
but I mean
°gosh°
They're dangerous too
10 and kids don't real-
His one friend
uhm
gosh he-
I said
->15 fl </wha:t /↑h:appened. < slow>
fl did a /tr:u:ck hit him?
?fl <the kid ('s got) his faxe, <listing>
?fl his who:le l:e:g,»
-> f2 «h+p> ↑well I was rollerblading;> <lax, breathy
20 - <and I'm going> <whispered>
fl <wooh!> <whispered>
A: - <yeah <whispered>
I mean>
B: [WELL ACTUALLY YOU SHOULD HAVE-
25 A: [you can't stop!
B: YOU SHOULD HAVE ELBOW PADS
KNEE PADS [AND A
W: [ « h > oh yeah!>
B: AND A HELMET ON.
30 A: you can't stop!
got) his face, his whole leg, with noticeable lengthenings and slow tempo.
These lines (17-18) are thus interpretable on one reading as Wanda the figure
making explicit to her interlocutor what the visual signs were which occasioned
the concern she expressed in lines 15-16.13
In line 19 a new figure enters.14 Despite the lack of any reportative
introduction, the presence of a different figure is clearly marked by a shift in
prosody: the line well I was rollerblading is configured with overall high pitch
and low volume. The voice quality is breathy and the articulation lax. It is the
prosodic contrast between this line and prior talk which cues the new figure.15
Yet not only the contrast between this line and prior talk is noteworthy,
but also the way the contrast is constructed. The contrastive figure being
enacted is a particular kid, with particular characteristics, ones which the
recipient-designed prosody of his voice cues: rather than the slow, weighty
delivery of lines 15-18 - cued as belonging to Wanda the figure - line 19 comes
across as light and airy. The kid is thus not only distinguished from Wanda, his
voice is also designed to suggest an opposing stance. Both sets of prosodic
features can be heard as cueing stances which are hinted at elsewhere in talk:
Wanda the figure's 'weighty' prosody becomes an index of they're dangerous
too (line 9), while the kid's 'light and airy' prosody indexes kids don't real(ise)
(line 10).
Notice now that the next two lines enact Wanda the figure's reaction to
the kid's response and its stance. Her wooh (line 21) is delivered in a whispered
voice, which 'leaks' into the prior reporting construction and I'm going (line
20). The whispered quality of Wanda the figure's turn here is significant in two
ways. First, it forces us to refine the statement made earlier that figures' voices
are animated consistently in conversation. If this were true here, Wanda the
figure should say w:ooh: and use normal volume just as she does in lines 15-18.
Instead, the whispery prosody of this line is designed to cue Wanda the figure's
reaction to the kid's response, and this stance is not - for want of a better label -
'serious concern' but something closer to 'speechless (or voiceless) amazement'
(cf. her gosh in lines 8 and 13, the first time also done with whispery prosody).
Thus, to have used the slow, rhythmic prosody of lines 15-18 would have cued
the wrong message. Yet although Wanda's voice is not consistently done, she is
constructed as a consistent figure, i.e. someone who is concerned by young
rollerbladers' accidents and amazed by their carefree attitude.16
Figures, we have stressed, are voiced as specific 'other voices': not only have
their 'lines' been constructed for them to 'say', their prosody hints at stances
26 Elizabeth Couper-Kuhlen
which they are constructed as assuming - and/or which their constructor, the
current speaker, wishes to take towards them. The speaker, in other words, is
not only 'doing' voices but also doing something with those voices which it is
incumbent upon recipients to infer. Every detail of the way figures in reported
speech sequences are constructed is thus inspectable for some clue as to what
the speaker is doing with these 'voices'. In particular, the way a given 'voice' is
configured prosodically and paralinguistically will serve as a hint to the 'take'
the speaker has on that figure, suggesting how it, or the talk of which it is part,
is to be received and evaluated by interlocutors.
How do interlocutors make sense of the prosodic and paralinguistic cues
of reported speech? Or better: how do we as analysts know what sense they
have made of them? What signs of participant 'coherence-making' are visible in
the interaction itself? One type of sign is verbal in nature: recipients may engage
in meta-talk which displays how they have interpreted the prosodic cueing of a
particular figure:
(6)
Breathing in smoke (12/106)
Same speaker constellation as in (3). The topic of conversation here is the ban on
smoking in public buildings which has been instated since Ann left the country to
study abroad.
Ann animates two figures here, both of which are herself at different
moments in time. The 'early' Ann figure is made to say lines which are hearable
as supportive of the anti-smoking ban actually you know I don't like to (..)
breathe in all this smoke all the time (lines 8-10), whereas the 'later' Ann figure
claims that those who are prevented from smoking should be able to do so (line
13), thereby implying that she no longer supports the ban. Both 'voices' are set
off prosodically from Ann's framing talk {it was just kinda like, line 7; and I'm
just kinda like, line 12) by pauses, pitch shifts and volume shifts. But the way
the 'later' Ann figure is voiced hints at something more: smoke in line 13 can be
heard as having the beginning of an overlaid 'snort' and the line what are you
guys doing, delivered with raised larynx, is followed by a laughter particle.
These delivery features are signs that Ann the narrator is not just contrasting an
earlier state of mind with a later state of mind on smoking bans but that she is
also evaluating these positions, aligning herself with the latter as opposed to the
former.17 Ann's interlocutors respond in ways which are hearable as aligning
with this critical stance. In fact Janet's next turn actually formulates verbally
what Ann was alluding to with her 'snort' and laugh particle: it's a little
overboard (line 17).18 Moreover, Ann ratifies this understanding with her
follow-up yeah (line 18). Thus, Janet's turn amounts to 'putting into so many
words' the effect of the work which prior speaker's vocal animation was
designed to do. It is metapragmatic discourse (Silverstein 1993) par excellence.
The Galileo fragment (4) also provides an example of metapragmatic
reference to the work which prosody is doing in a reported speech sequence.
In line 26 Steve responds to Janet's enactment of the Vatican
proclamation with that's the speed at which they uhm. On one level this
utterance is hearable as an explicit reference to - or formulation of - the point
which Janet is making with they just decided what, like last year (line 1f), they
just decided that (line 19) and five hundred years later (line 21). On another
level, however, Steve's remark can be heard as putting into so many words
what Janet's (slow and laborious) prosody is cueing in her animation of the
Vatican's voice. On this reading, it 'verbalises' the coherence which Steve is
attributing to the way the reported speech is done.
On other occasions, rather than verbalising the effect of the vocal framing
of a figure or figures, recipients will instead show their understanding of what
the speaker is doing by making responses tailored specifically to the reported
speech sequence. The Rollerblades fragment (5) also provides an example of
this.
Recall that Wanda the figure is animated with 'weighty' prosody, indexing
a stance which Wanda the narrator has articulated verbally with but I mean gosh
they 're dangerous too (lines 7-9), while the kid is animated with a breathy, light
voice, indexical of the stance implicit in Wanda the narrator's kids don't real-
28 Elizabeth Couper-Kuhlen
(ize) in line 10. In a sense then Wanda the narrator has 'primed' the vocal
characterisations of her figures and in doing so, set up two models for co-
alignment, one roughly associated with the danger motif, the other with not
realising the danger. Therefore, it is not wholly coincidental that her
interlocutors orient to these models in subsequent talk. Ann picks up the danger
motif by specifying why rollerblades are dangerous - you can't stop (line 2),
while Bret addresses the kid's non-realisation by detailing what protective
equipment rollerbladers should wear: actually you should have elbow pads,
knee pads and a helmet on (lines 24, 26f and 29). 19 Thus here participants show
through responses specifically designed with respect to the perceived stances
what understanding they have of the way the figures' voices are being done. In
this case, the inferencing job is made easier by the verbal hints which current
speaker has provided in prior talk.
Yet what if a current speaker's 'priming' talk is ambiguous? A final
fragment demonstrates that this can and does happen in conversational reported
speech sequences:
(7)
Rented cars with phones (12a/200)
Same speaker constellation as (5). Reference has just been made to cellular
phones.
20 sometime,
W: «p> hn>
last time,
W: «p> isn't that funny?>
OHHH
25 at- at Christmastime.
B: yeah,
W: right after Christmas.
yeah
B: he was so:: worried;
30 fl that his /mom was out in the /car,
fl and she had /rented one;
fl <that didn't have a /phone;> <rhythmic,
stylised melody>
C: - ha ha ha
B: fl <I hope she's all right; <rhythmic,
35 fl she can't /call us; same stylised
fl she /hasn't got a /phone;> melody>
W: heh heh
(he) takes after his father.
I don't worry about stuff like that.
40 they do though.
Bret's priming for the figure of Ricky, whom he animates in line 6ff,
becomes apparent for the first time in line 3: Ricky was so worried. This verbal
characterisation is repeated in line 17: he was so worried cuz- and once again in
line 29: he was so:: worried. Yet the way Bret 'does' Ricky's voice is not
indexical of worry throughout the animation. Line 6, what if she has a flat tire,
is configured with somewhat softer volume and lower pitch than prior talk, a
style of delivery which is indeed suggestive of worry. Yet in lines 7-8, although
Bret continues to animate Ricky's voice, the prosody changes gradually to
increasingly louder volume and marked rhythm; a high point is reached in lines
10-12, where the volume is very loud and the rhythm heavily marked. This
prosodic configuration suggests something more than worry, and indeed
recipients do not respond with, say, co-aligning expressions of concern and
sympathy but rather with giggling (line 9) and outright laughter (line 12).
Wanda even verbalises her understanding of Bret's voicing twice with isn 7 that
funny? (lines 16 and 23). Bret now animates Ricky once again, this time
employing, in addition to marked rhythm, a stylised melody ending with a call
contour (Ladd 1978) on each of the animated lines. This animation too is
receipted with laughter from recipients in lines 33 and 37. Thus, Bret has
30 Elizabeth Couper-Kuhlen
verbally cued Ricky as being worried, while his voicing of Ricky is suggestive of
something which provokes giggling and laughter in recipients.20
This example is telling because it reminds us that recipients do not rely
blindly on verbal 'priming' by the speaker in trying to make sense of the
prosodic and paralinguistic cueing of an animated figure in a reported speech
sequence. Instead, in cases of non-congruence, the verbal hints must be weighed
up against the prosodic hints and a more or less warranted 'guess' must be
hazarded as to what the speaker is doing. This is of course risky business, but
example (7) shows that it need not lead to a break-down of coherence.
4 Conclusion
In conclusion, the above discussion has shown that not only when explicit cues
to reported speech are lacking but also when explicit cues about reported
speech are misleading, participants make sense of conversational reported
speech in part by relying on the prosodie and paralinguistic details of 'voicing'.
As Schegloff reminds us, "talk is laced through and through with inexplicitness
and indexicality" and this inexplicitness is constantly being "solved" by hearers.
Moreover, "its results are displayed (even when not formulated) in the ensuing
talk and action and are subject to repair there if found problematic" (1996:
219f). This paper has attempted to spell out exactly what might be problematic
about indexically (or prosodically) cued reported speech, thereby making it
repairable, and how in more 'successful' instances, recipients display their
'solutions' of the inexplicitness involved to each other and to analysts.
Notes
1. The collection of articles in Auer and di Luzio (1992), takes one step in this direction.
2. This description is intended to be general enough to apply to all forms of reported
speech, including canonical direct as well as canonical indirect speech.
3. To keep the two notions apart, I shall use scare quotes around voice when reference it to
the animation of a figure; voice without scare quotes refers to phonatory voice.
4. I use the term reported speech sequence for any part of a conversational exchange in
which the presence of reported speech can be identified.
5. Transcription conventions:
One line One intonation phrase
First word capitalized High onset (=declination reset)
[Line
[Line Overlapped utterances
Coherent Voicing 31
Line=
=Line Latched utterances
Line. Final pitch falling to low
Line! Final pitch falling to low, emphatic
Line; Final pitch falling slightly
Line - Final level pitch
Line, Final pitch rising slightly
Line? Final pitch rising to high
«p> Line> Piano
<<ƒ>Line> Forte
« / > Line> Low register
«h> Line> High register
«cresc> Line> Crescendo
«decresc> Line> Decrescendo
ÎWord Noticeable step-up in pitch
Wo::rd Lengthened sound or syllable
Word- Cut-off sound or syllable
WORD Loud volume
°word° Soft volume
word (Extra) stress
/word /word /word Rhythmic delivery
(h) Breathiness
.hhh Inbreath
hhh Outbreath
(word) Unsure transcription
(.) Brief pause
(1.0) Measured pause
Which canonical form is being projected is unclear. The quotative he said is compatible
both with the presence of expressive elements (canonical direct reported speech) and
with their absence (canonical indirect reported speech), whereas the quotative he justified
it as, canonically speaking, projects upcoming talk in which expressive elements are
absent and would thus be considered indirect reported speech.
The noticeable pause following embryos? (line 28) is thus attributable to Gina. Her
silence here may be a first indication of the problem which Bill's conflicting signals are
creating for her.
Alternatively, the breathiness and laughter particles could be interpreted as indexing the
reporting situation, signalling the narrator's commentary on the figure's action (see
below).
Steve's laughter in line 16 can also be thought of as a sign of orientation to Janet's
'voicing' of the student.
10. The phenomenon of rhythmic scansion in everyday discourse is discussed at length in
Auer, Couper-Kuhlen & Mueller (forthcoming).
32 Elizabeth Couper-Kuhlen
11. In fact, it is partly due to the distinctive rhythmic cueing that we recognise the voice in
line 43 as the student's and not, for instance, the professor's.
12. Interestingly, the grammatical framing is that of indirect reported speech, not canonically
associated with 'expressivity' or vocal animation: they just decided (line 1) <...> that
(line 3) <...> that Galileo was right (lines 14ff), that maybe he had a good idea (line 25).
See Günthner (1997) for further discussion of this point.
13. Alternatively lines 17-18 could be attributed to Wanda the narrator, with their marked
prosody contextualising her 'take' on the events she is recounting (see below). In this
case the switch to Conversational Historical Present (Wolfson 1979, 1982; Schiffrin
1981) would begin here rather than in line 20.
14. Although we might expect Ricky to enter the scene at this point, the wording of the line
makes clear that Ricky's friend, the kid, is answering Wanda's question.
15. The fact that prior talk contained a first pair-part (what happened?) which makes a
second from some other (addressed) party conditionally relevant may also contribute to
the fact that we hear line 19 as belonging to a different figure.
16. In other words, as Tom Luckmann and Susanne Günthner (p.c.) have pointed out to me,
these characteristics do not contradict each other but are quite compatible.
17. The overlaid 'snort' and the laugh particle could be thought of as cueing the talk of
either the 'later' Ann figure or the narrator Ann. Since Ann the narrator 's stance is
congruent with that of Ann the 'later' figure, the net outcome is the same as far as the
inferencing here is concerned. See, however, fragment (7) below and its analysis.
18. In this sense it is the same practice as that documented in Schegloff's (1996) collection,
namely formulating a candidate observation, interpretation, or understanding of
something which a prior speaker has conveyed without saying: "[...] some telling may be
constructed by its teller, and/or be taken by its recipient, to embody and/or to reveal a
tack that the teller is taking to the tale, some stance being taken up, or some action being
done. When a recipient makes that explicit in the uptake, the teller can confirm both the
particulars of the uptake, its 'propositional content' so to speak, and that he or she was
engaged in such a 'project'" (1996: 188).
19. Note too that Ann 'chimes in' with Janet's vocal framing when she whispers her uptake
yeah I mean (line 221).
20. Note that the broken off I got a kick outa (line 1) and the expressive lengthening on so::
worried (line 29) both serve as cues to Bret's overall framing.
References
Auer, P. and Luzio, A. di. (eds). 1992. The Contextualization of Language.
Amsterdam: Benjamins.
Auer, P., Couper-Kuhlen, E. and Mueller, F. 1999. Language in Time: The Rhythm
and Tempo of Spoken Interaction. New York: Oxford University Press.
Couper-Kuhlen, E. 1993. English Speech Rhythm: Form and Function in Everyday
Verbal Interaction. Amsterdam: Benjamins.
Coherent Voicing 33
Ronald Geluykens
Westfälische Wilhelms- Universität Münster
1 Introduction
This paper approaches the notion of coherence from a perhaps somewhat
unusual perspective, viz. that of conversation-analysis (CA). I will argue here in
favour of a more empirically grounded approach to coherence. In fact, although
the notion of coherence is perhaps a useftil cover term for the kind of
phenomena which interest us here, giving a precise characterisation of the term
is not an objective of this paper. Interestingly, CA is defined in Crystal (1997:
92) as "a method of studying the sequential STRUCTURE and COHERENCE
in conversations" [my underscoring]. In the CA literature, however, the term
coherence is conspicuous by its absence. It is striking, for instance, that the
keyword coherence is not even listed in Harvey Sacks' lecture notes (1992).
However, in the local organisation of conversation, interlocutors appear to
reach agreement interactively on what they are talking about, and in doing so
they create coherence as they go along. What I will be mainly dealing with here,
then, is with topic organisation as a reflection of conversational coherence. We
will look at stretches of conversation in which (at least) two participants
attempt to come to some agreement on topical coherence by negotiating about
it.
It is important to point out that this methodology, and that of CA in
general, differs crucially from approaches to coherence in discourse analysis
(DA) (as developed in e.g. Brown and Yule 1983; Stubbs 1983). In the DA
view, coherence is present in the text, and the analyst is able to attempt to find it
by developing some theoretical framework combined with some suitable
heuristics. In other words, this is a top-down approach to coherence. In such
approaches, topic organisation is connected to the notion of aboutness, and the
36 Ronald Geluykens
The point is simply that topical coherence cannot be thought of as residing in some
independently calculable procedure for ascertaining (for example) shared reference
across utterances. Rather, topical coherence is something constructed across turns
by the collaboration of participants. What needs then to be studied is how potential
topics are introduced and collaboratively ratified, how they are marked as 'new',
'touched off, 'misplaced' and so on, how they are avoided or competed over and how
they are collaboratively closed down (Levinson 1983: 315).
(1)
- step 1 : topic-preparing sequence (with possible elicitor)
- step 2: speaker-turn containing first mention of topic
- step 3 : topic-negotiating sequence
called boundaried topic shift (Button and Casey 1984), but it is important to
realise that stepwise transition (Jefferson 1984) can also occur.
Secondly, the main impetus for the introduction of a new topic can come
from the speaker (i.e. the participant who first mentions the new topic) or from
the hearer (i.e. some other participant). In the latter case, the hearer will
typically use a topic initial elicitor (cf. Button and Casey 1984). The difference
between these two types of topic-introduction is reflected in my distinction
between topic-proposing on the one hand, and topic-eliciting on the other hand.
Thirdly, there is probably some correlation between the way a topic is
prepared on the one hand (step 1), and the way it is negotiated on the other
hand (step 3). While an investigation into this potential correlation lies outside
the scope of this paper, it seems a logical assumption to make. A topic which is
explicitly elicited by the hearer, for instance, may have no need for elaborate
negotiation; conversely, a topic which is introduced 'out of the blue', without
any link to prior topic(s), may require extensive negotiation.
Finally, the fact that topics need to be negotiated opens the possibility that
attempts to introduce a new topic may be unsuccessful. This is in fact a
relatively rare occurrence, since participants in informal conversations tend to
be co-operative and supportive in this respect (cf. Bublitz 1988 for an extensive
discussion).
A final aspect of the topic-introductions considered here that needs to be
pointed out is the following. In the CA tradition, little or no attention is paid to
the social roles of speaker and hearer. Such social positions may, in fact, have
some influence on the way participants use questions as topic-introducing
strategies. Factors such as relative power, solidarity, and even gender, may have
an influence on who does the topic-eliciting and/or -proposing in a given
context. My CA-based analysis disregards such potential variables.
Data employed here are from the conversational files of the Survey of
English Usage, based at University College London.2
2 Simple negotiations
2.1 Topic-proposing by current speaker
(2)
A: oh yes it is a household god of some sort isn't it I should think or is it a dancer I
don't know it's got an enormous belly —
B: ((m))-
Topical Coherence and Collaboration 39
(3)
1 A: (...) he'd probably have burst into tears *and* committed suicide in front
of me
2 B: *yeah*
3 A&B (-- laugh) -
4 A: (- laugh)
5 B: well. *how eventful*
6 A: *((m)) (- gasps)* yes . ((indeed)). so -
7 B: so what sort's thisflatyou're in at the moment
8 A: well it. [∂:m] do you remember Jackie .
9 B: yes . m
10 A: it's her flat.
(S.2.12.120.8)
All the material in turns 2 to 6 indicates a closing off of the initial topic of
the first turn, and in turn 7 speaker B proposes something entirely new. As we
will see later on (section 3), in quite a few cases the current speaker will pre-
signal the fact that he/she is about to propose such a new topic. This makes
sense, as it may facilitate the negotiation process, and avoid potential confusion
as regards topic flow.
40 Ronald Geluykens
Whereas the questions posed by the speakers during their conversation in the
preceding section constitute the second step in the negotiating process (i.e. the
actual first mention of the proposed topic), the questions discussed in this
section merely attempt to elicit a topic on the part of the second speaker, and
thus represent the topic-eliciting first stage in the process. A clear example is (4)
below:
(4)
A: oh fine was there anything else [∂:m]
B: so I just [d] I'd left some some records in Smart's room last
night which I was all panic-stricken about 'cos they're not mine
A: m
B: but I told Ned
(S.7.1.e5.6)
(4')
Stage 1 (A): topic-eliciting question
Stage 2 (B): topic-introduction (response)
Stage 3 (A): acceptance of proposed topic
(5)
A: good . [∂:m]. any more that we want to talk about the library —
CF: I *I ((mean some of the some of the books aren't there at all))*
A: *((I)) mean this is this is this is* . absolutely at the heart
of your teaching (...)
(S.3.3.40.5)
Topical Coherence and Collaboration 41
In this exchange, speaker A seeks to elicit a topic, but at the same time he
determines that the topic should be related to the library; the hearer (CF)
responds co-operatively by uttering the semantically related referent some of the
books.
Whereas truly open-ended wh-questions leave it more or less entirely up
to the hearer to pick a topic, a wh-question like (5) above restricts the number
of options the hearer can select from as a possible topic. Let us have a brief look
at another example:
(6)
A: but what was your conference about
B: oh it's about hydrology . ((and that sort of thing)) -
C: is that water
B: *water*
A: *be more* explicit that's just water to us -
B: [dm]
C: the science of water
B: hydrology is the science of- science of the quantity of water
(S.2.8.b10.7)
(7)
B: aren't you going to tell me what it is -
A: no -
42 Ronald Geluykens
B: anyway [∂:]what was I going to say . oh yes so what time are you coming
this afternoon
A: [dm] . [∂w∂] as we said about four o'clock
(S.7.2.b5.4)
(8)
A: (...) oh I must do that sometime - oh yes one thing too . [d:m] . are you at all
interested in coming to the B Minor Mass -
B: ((when is it)) the fifteenth of April
(S.7.3.f40.1)
(9)
C: well now Elsie follow it up this way do you visit your own mother -
(10)
B: right. *I'll . ((think)) about that*
A: *if you do let me know* . yeah - m ~ what else . haven't been up to Wales
again have you . or
B: [∂:m]I went up at Easter -
(S.7.3.f48.1)
In all four instances, the hearer provides some signal within the topic-
proposing turn that such a topic proposal is coming up. Since the next speaker
is forewarned, this may facilitate subsequent negotiation of the pre-announced
topic.
Whereas we find a high frequency of cues which signal the fact that the
current speaker is about to propose a new topic, this is not the only way in
which a topic shift can be announced. Rather than announcing the new topic,
the current speaker can also provide some cues which signal closing of previous
topic. This is the case in (11) and (12) below (closing cues highlighted in
boldface):
(11)
C: so you exposed it
A: I exposed it yes ((but)) [i?] but they can't get it back - . well I mustn't go on
boringly talking about me what are you doing Geoffrey -
B: [∂m] [∂:]much the same old things .
(S.1.9.99.7)
Topical Coherence and Collaboration 43
(12)
A: well that's Eileen's four . and we covered Margaret
what's her son doing now - ((Dan))
*Dan -
Dan*
*oh well Dan's doing quite* well (...)
(S.1.13.74.1)
In (11), the speaker explicitly signals the topic change coming up, by
indicating that the current topic (me) is finished; in (12), the speaker does so
more indirectly, by summarising the previously discussed topics.
Needless to say, both types of cue (pre-signalling of next topic and closing
of current topic) can be combined, resulting in explicit boundaries. Perhaps the
most elaborate instance of pre-signalling in the database may be found in
instance (13) below, since it combines a closing cue regarding the previous
topic with a pre-introductory cue regarding the proposed topic:
(13)
A: (...)" [∂:m] you're very kind old Sam — bless you well that finishes that. [∂:m]
. ((now)) what was the other thing I wanted to ask you . [ïi] is . is it this year
that \d:] Nightingale goes -
B: [∂:]no next year --
(S. 1.1.15.10)
In this instance, the speaker not only explicitly closes the previous topic
(cf. that finishes that), but also announces his new topic (cf. what was the other
thing...).
It is worth mentioning one example in which the topic-introducing wh-
question is, as it were, 'announced' by means of some explicit verbal material in
a previous turn rather than in the actual topic-proposing turn; consider (14):
(14)
C: I thought of a . new word that I. marked to discuss with you . at some time
B: yes
C: but I I'd better look up to see if it's in the dictionary first what about opt out
B: *yes*
C: *now* that must have a Latin root
B: m
C: I guess it must be something to do with the army (...)
(S.2.5.a61.3)
44 Ronald Geluykens
Two things are worth mentioning about the exchange in (14). First of all,
note the explicit introducing material in C' s first turn; this paves the way for his
subsequent wh-question, as it prepares the hearer (who acknowledges the
introductory clause!) that something is about to come up. Secondly, (14) is also
interesting in that it represents an interactional variant on exchanges such as (2)-
(3) above. Note that in this exchange, it is the speaker himself rather than the
hearer who establishes the referent; the hearer merely prompts the speaker to go
on about his newly proposed topic (the word opt out). We thus get the
following underlying process:
(14')
Stage 1 (C) introductory material
Stage 2 (B) acknowledgement of introductory material
Stage 3 (C) topic-proposing (Wh-phrase)
Stage 4 (B) acknowledgement of Wh-phrase
Stage 5 (C) elaboration on proposed topic
Stage 6 (B) acknowledgement signal
Stage 7 (C) further elaboration on proposed topic
4 Complex negotiations
In all the instances of topic-proposing mentioned so far, the topic-introducing
process is very smooth: negotiation of the new proposed topic takes place
through the adjacency pair sequence, after which speaker and hearer appear to
agree on the new topic. The same can be said about the topic-eliciting
exchanges: after the current speaker elicits a new topic (through the question),
the next speaker provides one (through the response), after which this
suggested topic is developed further. Such smooth negotiations, however, are
Topical Coherence and Collaboration 45
(15)
A: m - do you remember Terry Greenbridge .
B: Terry Greenbridge
A: he was
B: *yeah who was he*
A: *he used to be* master of Hereford College -
B: was he a theologian **((or someth)) yes**
A: **yes that's right that's** right. yes . well he he wrote and told me all about
this .(...)
(S.5.9.7.9)
(16)
C: have you tried the bookcase -
B: bookcase
C: in our room —
B: eh.
C: you know the big glass fronted bookcase -
B; *yeah*
C: *the* new one
B: oh that [∂.m] no as a matter of fact (...)
(S.7.1.a57.7)
(15')
Stage 1 (A) topic-introduction (Q)
Stage 2 (B) request for clarification
Stage 3 (A) clarification of referent
46 Ronald Geluykens
If one allows for the fact that stages 2 and 3 can be repeated (as is the
case in the examples above), this schema represents the majority of more
complex instances. Note that the topic in such cases can be eventually
developed by the current speaker, as in (15), or by the next speaker, as in (16).
Instance (17) below is slightly different, in that the next speaker
contributes actively to the identification process:
(17)
A: yes but ((Power)) can ask him if he sees him every day -
B: yeah but you need to know who he is 'cos he is much the more important man -
(noise)
A: true --
B; I just don't remember at all.
A: what about the others I ought to ring up the others ((didn't)) I . there ((was))
another one I sent it to . Wills —
B: oh the people we met first.
A: yes
B: the quite attractive young man and the older man .
A: yes ~
B: yes we ought to I think they might be quite reasonable - they looked a bit more
sort of ((rooking)) -
(S.4.2.10.2)
(18)
B what's that weird creature over there ~ *m*
C *in* the corner
B mhm
C [d] it's just a [dm] fern plant
Topical Coherence and Collaboration 47
In exchange (18), the identity of the NP that weird creature is not self-
evident from the extralinguistic context, which leads speaker C to assume
incorrectly that B means the fern plant rather than the television aerial that was
apparently intended. Once again, the conversation cannot proceed in a coherent
manner until this trouble spot is resolved.
Another instance of a complex interaction caused by the need for repair is
(19) below:
(19)
A: hm but they do disapprove of his [= McCarthy's] methods .
B: yeah most of them - yeah he has about twenty five per cent ((of)) support as
far as that's concerned and much more - not - and many don't know -
A: yes — what about in the university circles at Michigan —
B: at Michigan - *you mean Wisconsin*
A: *oh no no I mean* Wisconsin .
B: well I have never found anybody admitting to a liking for McCarthy or his
methods - *-*
A: *no*
B: . but nevertheless Gallup says . that in . university circles McCarthy is as
popular as he is in any other circles - ((2 or 3 sylls))
(S.2.1.b8.2)
5 Unsuccessful negotiations
In an earlier paper (Geluykens 1995), I have already devoted some attention to
what I called short-circuited topic-introductions. Since topic flow depends on
local negotiation through the turn taking system, successful introduction of a
new topic is by no means self-evident. In Geluykens (1995: 235-239), I
identified three underlying reasons which may cause short-circuiting. First of all,
the speaker may decide to abort a new topic in the middle of the negotiating
process; needless to say, this is a rare occurrence. Secondly, the hearer may
decide to interfere, and stop a topic from being developed (cf. below). Thirdly,
short-circuiting may be the result of competition for the floor, as appears to be
the case in (20):
(20)
C: m — and yet you feel terribly antisocial if you . ((you)) do just stay in the
kitchen anyway
a: yes - *what film*
C: *oh god those* stairs oh oh
a: - sorry
C: those stairs — **((you could . you could))**
a: __ **yes they're bad** aren't they - good exercise though
C: oh ((yes)) (- laughs) -
a: what film have you just been to see .
C: film.
a: I thought you went. you were going to the National - Theatre - National Film
Theatre
C: no no . [∂:m] . that was at the weekend (...)
(S.2.7.10.5)
(20')
Stage 1 (A): topic-introduction (first attempt)
Stage 2 (B): non-acknowledgement (side sequence)
Stage 3 (A): topic-introduction (second attempt)
Stage 4 (B): successful topical coherence
(21)
C: but are books more *expensive in Australia*
A: *tell them with my compliments*
all: (— laugh)
A: artificially cheap my arse (- laughs)
C: I say are they more expensive in *Australia*
A: *oh* god yes-(...)
(S. 1.10.97.5)
In this exchange, participant A has simply ignored the question the first
time, in other words he has not made any attempt at being co-operative; note in
this respect the overlap between the first two turns. This leads to the speaker to
repeat his question, resulting in a four-stage process similar to (20').
However, even without such floor-competition, topic-introductions may
be unsuccessful. In (22) below, for instance, the topic introduced by speaker B
is not responded to:
(22)
A: *m* [...] have a glass of sherry .
B: oh that's nice ((of you as)) I'm not driving . thank you
A: bloody hell —
B: but what about you \d] Crispin . [d] what's what's [thi] - how far were you
[∂:m]
((ye gods))
[∂:m] banking on *this*
*((...)) this k e y -
locked yourself out .
yes
(. laughs)
no the trouble is . oh for god's sake the key ((won't go in the lock)) (...)
(S. 1.2.52.4)
(23)
B: I'm afraid [thi:] fifth of February isn't all right for [thi:] - for [thi:] [dm] lunch
with Mr Parrot.
A: not
B: no cos he had something that he [?] . that I didn't realise .
A: [∂h∂ . ∂:m] well have you any other suggestions .
B: well he didn't give me any (- laughs)
A: ch
B: [∂:m] -
A: can I make . have you got his book there I mean
B: yes I have yes
(S.8.2.f3.2)
(24)
A: (...) I could go on being a research assistant going up and up and up until I
((was at)) . eighty five or whatever
C: m -
A: so what's new Ann --.
C: well I don't know if anything's terribly new at all . really . ((or is it)) all much
the same (. laughs) - *((just been))*
B: *((you still)) living* with Deb -
C: no no(...)
(S.2.7.48.8)
Topical Coherence and Collaboration 51
In this exchange, the next speaker again (C) explicitly states that he is not
really in a position to introduce a new topic; another participant then decides to
suggest a new topic (the living with Deb in the fourth turn), which the original
addressee replies to, thereby developing this new topic.
One final exchange containing a topic-eliciting question which is not
appropriately responded to is worth mentioning here, viz. (25):
(25)
A: (...)- you will certainly be involved I should think .
D: ah
A: because I was talking to them about their language policy over the next few
years
D: yes .
A: [∂:m] . yeah what [el] what other news have you got -
D: [∂:]
A: [so] I [wo] [d] I should just add that II meant to send you [∂:] a postcard
signed by Julius and myself but you know the way it is
D: yes
A: [w∂] you get chased around like a scalded cat (...)
(S.9.2.111.1)
In this instance, the original speaker (A) ends up responding to his own
topic-eliciting question. This appears to be due to the fact that next speaker D,
when given the floor, fails to come up with a timely response (cf. D's hesitation
in the sixth turn); speaker A then decides to resume the floor. Although this
exchange cannot really be labelled unsuccessful, it once again shows that, when
participants do not do the expected thing in terms of local topic organisation,
some re-organising is called for in order to preserve or create topical coherence.
6 Conclusion
I have tried to show in this paper how conversational coherence is
collaboratively achieved between the interlocutors. In particular, topic flow,
which is arguably one of the most important measures of a stretch of
conversation, depends for its regulation on the negotiation of new topics. It was
shown that the conversational activity of asking questions can be employed for
these purposes. Question-answer adjacency pairs can provide the framework
through which topical coherence is achieved, and they can do this in two
different ways. First of all, the current speaker can propose a new topic in
52 Ronald Geluykens
his/her question, giving the next speaker the opportunity to accept (or decline)
this proposed topic in the second part of the adjacency pair, the response to the
question. Alternatively, the current speaker can ask a question which elicits a
potential new topic from the next speaker. In both cases, we appear to be
dealing with boundaried topic shifts, i.e. some signal is given that the topic-
introducing question which is about to follow is unrelated to the previous
discourse, and/or some cues are provided that suggest that a previous topic has
been rounded off.
When such negotiation initially fails, participants strive to resolve this
potential trouble spot in subsequent turns. In that case, the potential outcome is
one of two possibilities. If the next speaker is unable to accept the proposed
new topic for some reason (e.g. because some element in the topic-proposing
question is unclear), then some further negotiation will take place until this
trouble spot is resolved. If, on the other hand, the current speaker's proposal of
a new topic is not accepted by the next speaker (e.g. if he/she has an alternative
topic they want to pursue), then the negotiation process will be suspended,
sometimes temporarily, sometimes permanently. In this case, too, participants
will collaborate until a satisfactory point is reached where coherence is again
apparent.
It would be unwise to think that this paper presents a complete picture of
conversational coherence. For one thing, one could question whether topical
coherence - which in itself depends to a substantial extent on referential
coherence - is a sufficient prerequisite for achieving conversational coherence.
For another, this paper is limited to one particular conversational device, viz.
questioning, as an illustration of the collaborative nature of coherence. Further
studies will hopefully provide us with a more complete picture of the variety of
strategies interlocutors use for creating topical coherence through the turn-
taking system (cf. Geluykens 1988, 1991, 1992, in prep.).
Notes
1. Thanks are due to Wolfram Bublitz, Uta Lenk, and Eija Ventola for their comments on
an earlier version of this paper.
2. Transcription conventions for the current purposes are the following:
—speaker identity: A, B, C,...
—overlapping speech :*...* or **...**
—dubious transcription or intranscribable speech: ((...))
—occurrence of pauses (from short to long):., -, - , —
—phonetic transcription: [...] (e.g. [d] = schwa)
Topical Coherence and Collaboration 53
References
Atkinson, J.M. and Heritage, J. (eds). 1984. Structures of Social Action: Studies in
Conversation Analysis. Cambridge: Cambridge University Press.
Bublitz, W. 1988. Supportive Fellow-Speakers and Cooperative Conversations.
Amsterdam: Benjamins.
Button, G. and Casey, N. 1984. "Generating topic: the use of topic initial elicitors". In
J.M. Atkinson and J. Heritage (eds), Structures of Social Action: Studies in
Conversation Analysis. Cambridge: Cambridge University Press, 167-190.
Crystal, D. 21997. A Dictionary of Linguistics and Phonetics. Oxford: Blackwell.
Geluykens, R. 1988. "The interactional nature of referent-introduction". Papers from
the 24th Regional Meeting of the Chicago Linguistic Society: 141-154.
Geluykens, R. 1991. "Topic management in conversational discourse: the collaborative
dimension". Papers from the 27th Regional Meeting of the Chicago Linguistic
Society: 182-195.
Geluykens, R. 1992. From Discourse Process to Grammatical Construction: On Left-
Dislocation in English. Amsterdam: Benjamins.
Geluykens, R. 1993. "Topic-introduction in English conversation". Transactions of the
Philological Society 91:189-214.
Geluykens, R. 1995. "On establishing reference in conversation". In R.A. Geiger (ed),
Reference in Multidisciplinary Perspective. Hildesheim: Olms, 230-240.
Geluykens, R. (in prep). "Information flow through interaction: Topic organization in
English conversation". Ms., University of Münster.
Jefferson, G. 1984. "On stepwise transition from talk about a trouble to inappropriately
next-positioned matters". In J.M. Atkinson and J. Heritage (eds), Structures of
Social Action: Studies in Conversation Analysis. Cambridge: Cambridge
University Press, 191-222.
Levinson, S. 1983. Pragmatics. Cambridge: Cambridge University Press.
Sacks, H. 1992. Lectures on conversation. (2 vols.) Oxford: Blackwell.
Sacks, H., Schegloff, E.A. and Jefferson, G. 1974. "A simplest systematics for the
organization of turn-taking in conversation". Language 50: 696-735.
Learning to Cohere: Causal Links in Native vs.
Non-Native Argumentative Writing
Gunter Lorenz
Universität Augsburg
case may differ across cultures and educational contexts, and there may likewise
be differences as to an argument's structural properties. Yet all four corpora are
meant to serve the same - argumentative - purpose.
Similarly, the connective signals under scrutiny are not defined formally,
but functionally: the corpus searches not only include the well-established causal
connectors, but also attempt to explore the more ad hoc type of causal linking
'strategies'. In this manner, the investigation tries to acknowledge that logical
relations are often lexically paraphrased, rather than being expressed as one
single conjunction, adverb etc. As paraphrasing strategies are known to be a
prominent feature of non-native discourse, a search which merely considered
the standard connectors would be severely invalidated.
Such a functional selection of lexico-grammatical data can, of course, not
be carried out in a purely mechanical way. Each potential item has to be viewed
and evaluated in context. Not all causal markers are causal in all their usages
(compare, most notably, since and for) and the numbers would be skewed if all
that was done were to count word-forms. Moreover, frequency profiles are not
the goal of this investigation, but rather its starting-point. Differences in number
may point to differences in style, which can only be properly gauged manually.
Seeing as — — 0.9 —
in that ... ... 1.9 20.0
in asfar as 0.7 — — —
that is why 13.4 11.7 5.6 LI
this is why 4.9 ... 3.8 2.1
which is why 0.7 ... 0.9 1.1
Σ (SF) 443.2 282.8 321.4 398.1
low-frequency items. The figures for because point towards a 'wild-card use'.
Secondly, there is a distinct decrease of because from BWF to UNI, i.e. from
the less to the more mature learners. At face value, this development is a very
welcome one, and one which appears to be set to rectify the statistical learners'
overuse. On further inspection, however, the learners' decreasing use of
because is more plausibly interpreted as part of an overall trend: as, since and
for are also declining from BWF to UNI, while undergoing a marked rise from
GCE to LOC. Without pre-empting a more thorough evaluation, there should
be a stylistic explanation to this numerical pattern.
The second group, on the other hand, does not really offer many insights;
the items are comparatively infrequent - with one small exception: explicatory in
that is a markedly native usage, mostly occurring in LOC, the more mature
corpus.
Group three conjunctions, while being only marginally more frequent, are
nonetheless slightly more revealing: the learners employed more than twice as
many paraphrases of the "pron is why" type than the native writers (30.7 vs.
14.6 SF). This finding confirms the suspicion that learners are more prone to
using explicit paraphrasing strategies. (Note also how in the native speakers'
data, too, such constructions occur much more frequently with the younger
population, indicating a negative correlation with linguistic maturity.)
It can already be seen how the developmental corpus structure enables a
kind of reasoning that a simple NS/NNS comparison could not achieve. In order
to substantiate the findings, the most frequent causal conjunctions will now be
subjected to a more qualitative analysis.
may disregard target-language restrictions which are not that obvious, or even
accounted for in the standard grammars, but which are nevertheless observed by
the native speakers. Such 'simplification' is one of the most frequently cited
features of learner language.
In an attempt to uncover instances of simplification in the present learner
data, causal because was carefully re-evaluated. This analysis revealed an
interesting pattern in the conjunction's position: it is only in the learner corpora
that because (excluding prepositional because of) appears in sentence-initial
position (BWF 9.2, UNI 10.3 SF). The NNS usages with sentence-initial
because fall into two categories; here is a sample of the first one:
01 Because I'm very interested in that sort of job, I'd like to apply for it. BWF
02 Because I couldn't get the permission to visit my aunt at that late point of time
I had to drive home again which took me another six hours. UNI
03 Because they can't find a solution they form two smaller groups. BWF
04 There are millions of possibilities to follow a new trend and therein lies the
difficulty. Because everybody has to have a trend to be possessed with,
otherwise one doesn't fit into our up-to-date society. UNI
05 If you keep this in your minds you don't have to be ashamed if compared with
European students. Because you can use your reason to see the things all men
have in common. BWF
06 All countries on the whole world have only one environment and it's up to us,
to everybody of us, to help it to survive. Because if the environment doesn't
function any longer the exitus of men is just a matter of time. UNI
Unlike the first set, these examples do follow the unmarked sequence of
"EFFECT→CAUSE", but they do so in a fashion unusual for this type of
discourse. By breaking the sequence up into two sentences, the concept of
CAUSE appears to be added as if in an afterthought - which would more likely
occur in spontaneous speech than in formal writing. This interpretation, albeit
informed by merely a small section of learner usage, adds another dimension to
the data: NNS style, as evident in the use of because, is not only governed by
wild-card use, but also by register-mixing, i.e. by the interference of other, less
formal text-types.
62 Gunter Lorenz
07 Many travellers are really interested in learning new things and understanding
foreign cultures. That's why travel also helps to connect people from different
backgrounds. BWF
08 It is not great fun to watch a movie that is interrupted by commercials every 15
minutes. That's why I think television commercials should be banned. UNI
09 First of all, I want to say that politics, as it is something made by man, can't be
perfect! That's why it is no use to tell you that this guy and that person have
done something wrong. BWF
10 Today many people have to live with haste and excitement, they have a hard
job, they do not earn much money, so it is understandable that they do not like
it even to work at home. BWF
11 But I think not only history is important but also the present time. So in my
opinion novels of modern writers should also be included in the curriculum.
BWF
12 However, no war is the same, as they have all different causes and objectives.
So, one should ask oneself the question, whether a war is justified or whether
peace is a good thing at any price. UNI
text-type sensitivity does indeed lie at the heart of the NS/NNS numerical
contrast.
13 Furthermore, let me add the fact that mass tourism leads to more understanding
for strangers on one's own country. BWF
14 A very high percentage of all accidents is caused by really unreasonable
drivers, who feel much too good in their cars. UNI
15 A really visible reason for the emancipated woman being alive is the high rate
of unmarried or divorced women. UNI
16 By watching frequently such discussions the citizens get to know the principle
aims and the guidelines of each party and as a consequence they can take part
in elections in a reasonable way. BWF
This small sample already shows that the learners use lexical causal
strategies in various ways. And despite the fact that quotes 13 to 16 contain
several instances of non-nativeness, the causal markers themselves hardly
appear to be chosen as mere substitutes for grammaticalised connectors. In
many ways, patterns such as "A leads to B" or "C is caused by D" connect
CAUSE and EFFECT in a far more direct and literal manner than an adverb or
conjunction would, for example. The same is true for explicit labellings such as
reason or consequence. It would therefore be imprudent to expect that lexical
markers should turn out to be an exclusively non-native domain. On the
contrary, as such literal reasoning puts more emphasis on the causal relationship
itself, native writers may well make copious use of it in the context of
persuasive argument. Compare:
17 The British tend to be very proud and possessive of the law, and changes may
well cause illfeeling and discontent. LOC
18 Three have died since 1986 in Britain alone but these deaths may have been
caused by "violation of the rules". GCE
19 The reason for such a large sixth form is our "open entry policy, which gives
anyone with the right attitude a chance to do A-levels". GCE
20 The consequence for business is that it has had to adapt its production away
from baby and children's products, to those required by the older generations,
in particular medical supplies. LOC
The corpus searches for the next section draw on the hypothesis that
lexical causal patterns will inevitably contain a variant of the lexical field of
CAUSE or EFFECT. Retrieval was therefore effected with the help of dictionaries
and thesauruses. As in the case of grammaticalised items, all elements have been
viewed and evaluated in context.
comply with the general tendency. Unlike causal so above, this is not likely to
be a question of text-type or register, but rather of another wild-card overuse on
the older learners' part.
As we have seen in the case of because, non-native wild-card use may
well lead to the violation of target-language regularities. It is therefore not very
surprising that the same was true for the learners' use of cause. To understand
the nature of the NNS irregularity in question, we must take recourse to a
concept which has recently become associated with the term 'semantic
prosody'. It has repeatedly been noted that cause - despite its seemingly
neutral denotation - has a certain negative undertone and a strong association
with undesirable events or states of affairs. Native speakers are much more
likely to say that X caused a riot than to say that ?Y caused peace and quiet. In
cases where apparently neutral words systematically cooccur with markedly
positive or negative concepts, we speak of a positive or negative semantic
prosody.
The negative semantic prosody of cause has been validated by a 120
million word corpus (Stubbs 1995). Yet even the present, comparatively small
corpora produce unequivocal results: the most frequent noun cause collocates,
for example, are damage [10 occurrences], problem(s) [9], accident(s) [7], and
death(s) [7] - doubtless all unpleasant concepts. In the few cases where the
collocates carry positive connotations (joy, gain) or neutral ones (reaction,
society), the context makes it clear that the negative bias still holds:
24 What does the American Dream mean for you? Is it only the cause of your
comfortable existence, the reason why you can feel pride in your country?
BWF
25 I will cause the rise of human power and dignity by giving man a help. An
inferior. No human intelligence, feeling or creativity will ever be replaced by
any kind of circuitry. BWF
26 Nowadays women without a husband very often cause respect, especially if
they have a good position in their job. UNI
As with causal verbs, the noun figures confirm that lexical causal patterns
cannot compensate for the NNS/NS differences in grammaticalised items - at
least not in terms of frequency. As the NS increase shows, causal labelling is not
an indicator of linguistic immaturity; on the contrary, the counts for the native
speakers continue to rise with age, and the increase is even stronger than before.
The learners' frequency counts show a slight rise from BWF (134.4 SF)
to UNI (137.5 SF), which is, however, too low to offer any compelling insights.
The NNS total is, for once, higher than the NS one (271.9 vs. 221.2 SF), and
again this is due to a marked overuse of the most frequent element. Among the
eight nouns under scrutiny, reason not only has the highest counts, but is also
Learning to Cohere 69
an exception to the overall trend: its numbers show a substantial non-native rise,
and a small drop from GCE to LOC. As the word itself is not marked for
register, the figures suggest another instance of NNS wild-card use. As such, it
is - naturally - far less needed on the native speakers' part, but increasingly
relied on in learners' writing.
Finally, and in keeping with the line of inquiry so far, reason is checked
for patterns of non-native usage. Two different ones emerge:
influence the sum totals (S3) significantly: the overall tendency remains the same
as outlined for Σ1.
If we can indeed conclude from the figures that causal marking positively
correlates with mature argumentative style, as hinted above, then this has
serious ramifications for the interpretation of the learners' developmental
decline. In simple terms, it would imply that the older learners' writing must
either be considered less mature, or "less argumentative" than that of their
younger counterparts. The former of these conclusions is not very likely; if it
were true, this would mean that the learning and teaching efforts between the
two stages had been counter-productive - at least for the small section of causal
marking. Fortunately, the second surmise appears to be far more plausible. The
high UNI figures for causal so and that's why indicate a manifest change in the
learners' style of writing: when asked to argue a case, the older learners adopt a
more casual, conversational register, shifting away from the more formal
school-type written argument. This shift of style, however, is not consistent; as
some features of formal writing are still prominent - see the high counts for
prepositional markers - it makes for a striking mix of registers.
In trying to explain the differences in native and non-native stylistic
development, we should consider the different points of departure. Whereas
adolescent native writers naturally have no problems with casual English
idiomaticity, it is their written register which is still developing. Advanced
German learners of English, on the other hand, are used to text-based argument,
modelled on the style of quality weeklies. In attempting to emulate such
journalistic prose, they resort to the structural signals typical of that type of
discourse - without, of course, having the corresponding lexical means and
argumentative skills. If the younger learners therefore exhibit a high number of
causal connectives, higher even than their native-speaking counterparts, this
may first of all be due to the educational fallacy of trying to get adolescents to
write like professional journalists - in a foreign language.
This puts the older learners' data in perspective: university students of
English are, of course, far more prone to register-mixing than school pupils,
simply because they have gradually become exposed to a wide variety of written
and spoken styles. The seeming deterioration in the learners' development is
hence most plausibly interpreted as the deconstruction of a didactic
misconception. The older learners' interlanguage contains both elements of
casual speech and formal writing, but the learners lack the intuitions which
guide a native speaker's usage.
The pedagogical consequence of this study therefore cannot be a mere
quantitative one: teachers would be ill-advised to simply try and increase the
number of connective devices in their students' writing. There is, regrettably, no
quick access to linguistic intuitions, particularly so in the finer points of usage,
72 Gunter Lorenz
Notes
1. The adjective causal is used here to refer to items which mark either CAUSE or EFFECT; in
either case, the logical relationship is clearly a causal one.
2. The four sub-corpora are part of an ongoing learner corpus project conducted by the
author; for more details see also Lorenz (to appear). While the corpus is still growing in
size, the word counts given here represent the stage of corpus search for this
investigation (early 1997).
3. No instance of in so far as or insofar as found in the corpus.
4. Note how this corresponds with the research findings quoted above. Within the
International Corpus of Learner English project, highly overused high-frequency items
have come to be known as 'lexical teddy-bears' - comfortably familiar and easy to cling
to. The term was coined by Angela Hasselgren, who has worked with Norwegian
learners' data at Bergen.
5. There was one exclusively native pattern which resembles that of quotes 04 to 06: where
the native speakers felt the need to add a reason in a single clause, they introduced it by
this is because [15 occurrences], thereby creating a full-fledged main clause, as opposed
to the learners' stand-alone adverbial one. Once more, it is surprising that the learners
did not resort to that pattern once.
6. Compare Schleppegrell (1996a, 1996b), who asserts that the use of because as such is far
more common in spoken than in written discourse.
7. For the purpose of this paper, such stylistic ratings are based on the information given in
monolingual dictionaries and usage-oriented handbooks (see references).
8. This term was coined by John Sinclair; see (1991:109ff) for 'open-choice' versus 'idiom'
principle.
9. The concept of textual 'label' has been borrowed from Francis (1986, 1994), the idea
being that a certain class of nouns can be used as identifying tags for preceding or
ensuing stretches of discourse. These tags can work within or across sentence boundaries.
In the first sentence of this footnote, for example, the noun idea can be seen as a label for
the whole ensuing that-clause.
10. See, e.g., Louw (1993), Stubbs (1995), Bublitz (1996).
11. Compare Ventola (1996), as well as Lorenz (1998).
12. It should be added here that in two cases the native speakers did employ a parallel
construction - namely It is for this reason that (both LOC). This is similar to what was
noted in footnote 5: two separate patterns, an NNS and an NS one. And as with This is
because, where the native speakers used a more elaborate focusing pattern, the learners
74 Gunter Lorenz
References
Bublitz, W. 1996. "Semantic prosody and cohesive company: somewhat predictable".
Leuvense Bijdragen 85: 1-32.
Connor, U. 1984. "A study of cohesion and coherence in English as a second language
students' writing". Papers in Linguistics 17: 301-316.
Crewe, W.J. 1990. 'The illogic of logical connectives". ELT Journal 44: 316-325.
Ehrlich, S. 1988. "Cohesive devices and discourse competence". World Englishes 7:
111-118.
Evenson, L.S. and Lintermann-Rygh, I. 1988. "Connecting L1 and FL in discourse-
level performance analysis". Papers and Studies in Contrasfive Linguistics 22:
133-178.
Field, Y. and Yip, L. 1992. "A comparison of internal cohesive conjunction in the
English essay writing of Cantonese speakers and native speakers of English".
RELC Journal 23: 15-28.
Francis, G. 1986. Anaphoric Nouns. Birmingham: University of Birmingham.
Francis, G. 1994. "Labelling discourse: an aspect of nominal group lexical cohesion".
In M. Coulthard (ed), Advances in Written Text Analysis. London: Routledge,
83-101.
Granger, S. and Tyson, S. 1996. "Connector usage in the English essay writing of
native and non-native EFL speakers of English". World Englishes 15: 17-27.
Halliday, M.A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Kennedy, G. 1992. "Preferred ways of putting things with implications for language
teaching". In J. Svartvik (ed), Directions in Corpus Linguistics. Berlin: de
Gruyter, 335-378.
Lintermann-Rygh, I. 1985. "Connector density - an indicator of essay quality?" Text 5:
347-357.
Lorenz, G. 1998. "Overstatement in advanced learners' writing: stylistic aspects of
adjective intensification". In S. Granger (ed), Learner English on Computer.
London: Addison Wesley Longman, 53-66.
Lorenz, G. 1999. Adjective Intensification - Learners versus Native Speakers: A
Corpus Study of Argumentative Writing. Amsterdam: Rodopi.
Louw, B. 1993. "Irony in the text or insincerity in the writer? The diagnostic potential
of semantic prosodies". In M. Baker, G. Francis and E. Tognini-Bonelli (eds),
Text and Technology. In Honour of John Sinclair. Amsterdam: Benjamins, 157-
175.
Milton, J. and Tsang, E. 1993. "A corpus-based study of logical connectors in EFL
students' writing". In R. Pemberton and E. Tsang (eds), Studies in Lexis. Hong
Kong: HKUST, 215-246.
Learning to Cohere 75
1 Aim
The purpose of this paper is threefold. I will begin by introducing the notion
discourse pattern as a technical term, relating it to cognitive aspects of
language use. Next, I will show how coherence can be viewed in terms of this
notion, and de facto in terms of understanding in general. Finally, I want to
illustrate the usefulness of the notion discourse pattern in discourse analysis at
large, through an analysis of news reports in a number of US and UK
newspapers on the meeting between presidents Clinton and Yeltsin in Helsinki
in March, 1997.
Both genre and text type are clearly important notions, as is the need to
differentiate between the two. Recent work both in linguistics and in literary
studies has indicated the complexity of the notions genre and text type; e.g.,
they have to be related to social practices, to intertextuality, and to the
processual construal of discourse. Nevertheless, genre and text type are very
often seen to form a dichotomy, as being two perspectives on discourse: genre
zooms in on the external relations that a text/discourse displays, and text type
focuses on its internal relations.
It is worth noting that when these two approaches are taken
simultaneously, they are, metaphorically speaking, supposed to meet half-way,
and in this manner circumscribe and define das Ding an sich - the coherence -
of the text/discourse. When the starting point for an analysis of a text/discourse
is a categorization in terms of a particular genre (machine-readable corpora like
the LOB and Brown corpora are typically organized in terms of genre), the
next step is to minutely describe the kinds of structures, phrases and words that
can be found in each genre - both very locally, within sentences and clauses,
and more globally, e.g., with respect to how texts of a particular genre start or
end. On the basis of such analyses we can then find linguistic similarities and
differences between the various genres that we had set up on an intuitive basis
(but which, indeed, have been codified gradually through socialization). Thus,
we get information about the manner, degree and range in accordance with
which discourse, and consequently coherence, can vary.
However, when the starting point - and initial focus of interest - is, say,
a narrative text type, linguistically defined as, among other things, having its
foregrounded units (sentences, utterances, information units) being
sequentially ordered so as to correspond to the order in which the events
described took place in 'real life,' the logical concomitant step is to specify the
differing characteristics of this particular text type as against other text types.
The typical next step is (cf. e.g. Werlich 1976) to generalize from here:
grouping together different kinds of narratives, one eventually ends up with
what Werlich called text form variants - which, for all intents and purposes,
are what others call genres; e.g., a fairy tale is a narrative text form variant.
Schematically, the two approaches can be displayed as in tables 1 and 2.
Genre and text type differences clearly need to be taken into account in
any linguistic analysis, e.g., analyses of language change is extremely sensible
to genre and text type differences. Still, despite the fact that genre studies and
text type studies have a very similar concern, it is also important to keep in
mind that genres and text types are not the same. A genre like 'novel' can be
written in the form of a narrative text type, but it can also be in the form of,
say, an argumentative or exploratory text type. A novel like Julian Barnes's
Coherence and Discourse Patterns 79
initially
construes its own context, of how discourse not only reflects, but creates
situations, a definition of discourse and its coherence cannot be adequately
given in the traditional terms presented above.
form/internal >
< function/external
The picture in figure 1 in fact implies that there are two toolboxes needed
for linguistic analysis. My view is, however, that these two complementary
approaches are not enough. We need a third toolbox. In very much the same
way as I think we need a 'meaning' filter 'in-between' form and function, we
also need a filter 'in-between' genre and text-type descriptions. This third
toolbox I will call a discourse pattern description.
There are several, even purely heuristic reasons for recognizing such a
level of analysis. For instance, I would claim that it is not enough to
characterize a newspaper report (= a genre description) as a narrative text type,
since it is doubtful whether readers and interactants understand a news report
primarily in such terms. What analyses in terms of genre and text type lack is
reference to cognitive aspects like understanding. Schematically, the view
advocated in the present paper can be presented as a three-way, rather than as a
two-way division where 'discourse pattern' mediates between 'text type' and
'genre' in a parallel fashion to the way semantics can be said to mediate
between Syntax and Pragmatics:
12x fdjkfjfd
23nn lkfdlkjf sdjklfdsjkl
34fr kjdfsjklfdiop
asälkfj oksd jfdkfj dfjfj dsklfj dfjsd fjsdfjfjf
söä df kskflsdkfd kf kdsfkdsf kdfkd fkdfkdsfl
öls kdflösd kfsdl kfsfksfk s dlfk sdfksäd fsdl
82 Jan-Ola Östman
Ingredients
Preparation
Text 1.
receives coherence, and in terms of which texts and discourse are readily
interpretable and understandable (and regularly interpreted and understood) as
being of that particular kind.
If coherence is seen in relation to cognition and understanding, in terms
of how particular interactants grasp and deal with discourse, the status of the
universality of coherence becomes an issue. In the case of the Recipe pattern,
we have a pattern that is very frequent in very many literate cultures. However,
if we look at another frequently encountered pattern, for instance, that of Death
notices, we find at once that any attempt at a 'universal' schematization of
these will be extremely different. Even neighboring communities and cultures
with close contacts use very differently codified Death notice patterns.
For instance, whereas in Britain, deaths are recorded as text, in Finland
deaths are advertised much more graphically. Still, both are patterned. Fries
(1990) has shown that British death notices like those under (1) have a typical
structure with particular slots of information in a particular order, cf. figure 4,
where both the general and a more extensive structure is depicted.
Died a few days ago, at his house in Greenwich, Capt. Robert Walter, of the Royal
Navy.
In Dublin, the Honourable Miss Isabella Howard, second daughter to the Right
Hon. Lord Clonmore.
Text 2.
Rakkaamme, insinööri
Pasi Antero
Ailio
s. 9. 7. 1951 Kajaanissa
k. 7. 10. 1996 Naantalissa vaikean
sairauden murtamana
Kaipaamaan jäivät
Marja-Helena. Heini ja Henri
Lea-āitî
Muut sukulaiset ja lukuisat ystâvàt
A Finn might not off-hand recognize, nor even understand or believe that
a British death notice is a proper death notice. For a death notice to be coherent
as a death notice to him or her it would need to have the structure, the pattern,
of a Finnish death notice.
The Recipe pattern and the Death notice patterns are strictly codified
patterns - relative to particular cultures. I want to suggest, though, that the
usefulness of the notion of discourse pattern for understanding coherence is not
restricted to formulaic texts.
One very pertinent question that needs to be raised is what linguistic
evidence there is for suggesting that we need the concept discourse pattern. I
will show this by providing minimal pairs in a tabular form (cf. table 4), and
indicating that the notions genre and text type are not enough.
The examples in table 4 indicate, first of all, that there is no one-to-one
relationship between genres and text types: both fairy tales and novels often
utilize a narrative text type; a novel can be in narrative form, or it can, e.g., be
an epistolary novel in an instructive format.
86 Jan-Ola Östman
The three first text examples - rows - in table 4 are all specified as
'Aesthetic' under discourse pattern. Without going into details about this
pattern here4, suffice it to say that it is built on the notion of the Golden
Section, which is a very prevalent structure in (traditional) fiction.
The last three examples in table 4 all utilize the descriptive text type, and
are all contact ads as to their genre. Patternwise, however, contact ads in
Finland (FIN) are very different from both British and American contact ads.
Finnish contact ads prototypically have the same structure and form as
ordinary advertisements for cars, washing powder or sausages: they all utilize
something to attract the eye, something that stands out from the background
and shouts to the potential reader: READ ME!. This pattern is depicted in
figure 5a as a contrast between (the salient) black and (the backgrounded)
white. Finnish contact ads typically contain phrases like HEY YOU! or
SPORTY BRUNETTE - utilizing, precisely, capital letters, italics, bold face,
etc. - and in addition, of course, they contain a description of who is writing,
and what kind of relationship and person he or she is interested in.
English contact ads are of two types, but very seldom of the Ad pattern
type of figure 5a. One way is to write them in the form of a story with a plot, a
clear beginning and end, very much like an ordinary description of a landscape,
but with a name, pseudonym, or telephone number at the end as the climax.
This pattern is depicted in figure 5b. It is the general pattern for an ordinary
story, with the internal boxes referring to the obligatory (more than one)
foregrounded parts of a story. Thus, as we see, when a visual picture is to be
taken as a graphic representation of how text/discourse is conceptualized, this
representation is also a visual abbreviation for what it means linguistically for a
text to be of this kind: The representation in figure 5b is a short-hand for all the
linguistic manifestations that partake in defining grounding in discourse.
Coherence and Discourse Patterns 87
n/x
All three (cf. figures 5a, 5b, 5c) are contact ads, all are descriptive (of
who is writing and who is wanted), but they are presented - and conceptualized
- in the form of different discourse patterns.
The use in different cultures of different discourse patterns can also tell
us something about how that culture relates to particular activities. For
instance, with respect to the suggested discourse patterns for contact ads, we
might want to propose that the way English contact ads (in both patterns) are
written suggests that readers are prototypically expected to read the contact ads
section as a whole, as a smörgåsbord, from which they then choose their
potential partner. Finnish contact ads, however, are like small news stories, like
ads, which prototypical readers might (or might not) browse through, and they
will not stop to read a particular (contact) ad unless drawn to it by a particular
READ ME! sign.
The usefulness of the notion discourse pattern for understanding
coherence should by now be obvious. But what kind of phenomenon is a
discourse pattern?
mnemonic devices, and the important thing is not what they look like per se,
but what they look like in relation to other figures. This holds true for
discourse patterns, too.
AAAA
®
Figure 6b. Galtung 's academic styles.
Top left: Saxonic (non-dialectic, thesis oriented); down left: Teutonic (non-
dialectic, theory oriented); top right: Nipponic (dialectic, thesis oriented);
down right: Gallic (dialectic, theory oriented).
writing, formal vs. informal, planned vs. unplanned, fictive vs. non-fictive,
prose vs. poetry vs. drama? It is obvious that the more all-embracing definition
a discourse pattern gets, the more abstract and emptier it becomes: soon
everything is discourse patterns - but thereby the concept also loses its strength
as an analytic device. My view is that like other images, discourse patterns,
too, have to be placed among the basic-level terms of cognition. This is the
level they are most useful for; they are needed as tools for investigating how
interactants understand coherence in discourse. It is precisely at the level of
discourse patterning that an audience recognizes that a text is, say, a death
notice. This is thus also the level at which a text has the most direct effect on a
reader or audience.
As I have argued, discourse patterns pertain to the holistic perception of
text/discourse; discourse has a particular structure which is associated with a
particular function and contents. That is, they are not simply shape, but they
function as frames for understanding. They are not only linked to perception,
but to cognition: they activate scripts, and thus function as guides for
understanding. Discourse patterns are directly associated with coherence in
terms of understanding. But if that is the case, it also suggests that discourse
pattern similarity implies similarity in manner of understanding, and similarity
in how we perceive and process texts.6 For instance, procedural types of
discourse like the recipe, the guide-book, and direction-giving as interaction,
have a very similar structure: first a presentation of the ingredients (Recipe),
the (history of the) places worth seeing (Guide-book), and the joint
establishment of mutually known landmarks and means of transportation
(Direction giving); then a step-by-step account of the process by which one
gets from ingredients to the finished product, or from point A to point B.
It is unclear at this point to what extent the Recipe, Guide-book, and
Direction-giving patterns should be conflated into one pattern, i.e., to what
extent the Recipe pattern of figure 3 could as such be applied to them all. There
are both good and not so good sides to such a decision: on the one hand,
assigning them the same discourse pattern would stress their similarity; on the
other hand, if a particular discourse pattern gets to be too abstract, it is no
longer usable as a basic-level term. In the present case, if these three were
conflated, it could then be argued that the joint pattern needs to be renamed as
something like the Procedural, or Incremental pattern. But if so, we are very
close to talking about text types, rather than discourse patterns, and thus to
losing the insights the latter notion provides.
Since what patterns look like in relation to other patterns is what is
representationally important, the notion of discourse-pattern similarity is by
definition a strong device in the analysis of textual/discoursal coherence. Thus,
the task of assigning a discourse pattern specification to a (sub)genre can
Coherence and Discourse Patterns 91
virtually any place and give space for the arrival of an important last-minute
news item.7
The picture that has emerged as an upside-down triangle is for all intents
and purposes also a feasible way to present the discourse pattern of traditional
news items.
There is, however, also another discourse pattern in news reporting that
has come to compete with the traditional triangular pattern. This is the Human
interest discourse pattern introduced in figure 5c, which visually presents a
story as containing several golden nuggets throughout the news report. This is
a very different kind of pattern. There is no way one can cut off at will a news
item that has been written using the Human interest pattern, because, as we
saw earlier, what is very typical of this pattern is that it has one of its largest
golden nuggets at the end of the report.
This human interest manner of writing news items has for long been the
typical way for evening, tabloid papers to tell an interesting story - very often
about people. British tabloids like The Sun and the Daily Mail virtually neglect
topics like the Helsinki Summit. But if they do mention such topics, they dress
them up in a human interest fashion. Thus, The Sun's reporting on the Helsinki
Summit was confined to a brief item hid away on p. 7 of the paper with the
headline Boris fury at 'threat by NATO' (March 22, 1997), concentrating on
Yeltsin's feelings: Mr. Yeltsin called the plan "unacceptable" ... "It is a
serious mistake ".
In Great Britain8 this division of having different discourse patterns
being associated with different kinds of newspapers still functions pretty well:
the triangular News pattern is dominant in news reports in broad-sheets, the
Human interest pattern in tabloids. However, in the United States, tabloid
newspapers do not have the same prominent status for the everyday reader as
in Great Britain. There are a number of tabloids in the USA, too, like the
National Enquirer, but typically the 'soft' news reporting of British tabloids is
to be found in commonly read weekly magazines like People and TV-Guide in
the United States.
Coherence and Discourse Patterns 93
The LA Times of March 22, 1997 starts by treating all the four basic
issues of the Helsinki Summit (NATO; Arms control; G7; Bill & Boris) in a
front-page article entitled "U.S.-Russia Talks End in Arms Breakthrough". The
coverage is then split into two items (plus a "News Analysis" item) inside the
paper (pp. 18-19). One article deals with the NATO issue, and the other article
with the Arms control issue. Both of these articles gradually peter out with
respect to their informational content, and thus one interpretation of the
structure of these news items would indeed be to say that the triangular-shaped
discourse pattern (figure 7) has been followed in these reports.
However, a closer look at the reports indicate that, although both of them
have 'less important' information at the end, this 'less important' information
is not wholly of the 'supplying background' type, but it is, precisely, of the
human-interest kind. For instance, the Arms control article, "SUMMIT: Arms
Breakthrough at Talks", ends with an account of Clinton's problem as he had
not been able to sleep; there had seemingly been some disturbing noise in the
hotel. These final paragraphs of the news report are given in text sample 4
below (the low-case letters a-c have been added for ease of reference).
(a)
(b)
told Yeltsin that he had a hard time sleeping between midnight and 2 a.m. because
of a "loud thumping" from the ceiling above him, White House Press Secretary
Mike McCurry said later.
"Boris, I thought you had hired an extra-large Finn to stomp on my
(c)
roof," the president joked, and Yel
tsin laughed, according to McCurry.
Coherence and Discourse Patterns 95
McCurry said he was unable to determine the real source of the noise that
interrupted the president's sleep and left him red-eyed and tired-looking during the
news conference.
Text 4.
The (a) segment of text 4 shows how suddenly the break comes between
the spread-sheet kind of reporting on taxes and other business of political
importance, and Clinton's habit of playing the saxophone at some point in his
free time during official meetings. What follows from here onwards is - in a
discourse-pattern interpretation - one big golden nugget of human-interest
reporting at the end of the news item.
Section (a) of text 4 runs almost to the bottom of the page (p. 18). Below
it we find a two-column sized picture of the two presidents seemingly in a
good mood. The caption given to this picture is President Clinton shares a joke
as Russian leader adjusts headset. It is unlikely that the journalist or publisher
actually heard or otherwise knows what went on between the two presidents,
but the picture is clearly a picture of the human-interest type: 'sharing a joke,'
'doing something insignificant like adjusting one's headset,' 'laughing,' etc.
If news items are organized (partly) according to the Human interest
discourse pattern, readers are not expected to stop reading after the lead.
Readers also, of course, very soon learn that it pays to read the whole article,
since at the end one will get a 'lighter shade' of the topic.10 The traditional
manner of presentation would have been to publish a separate Human-interest
insert recounting the incident of Clinton's sleeping problems. Combining the
two stories which have different kinds of news value, the contents of both
stories change: the stress falls on the protagonists as being the same persons in
private and in public, rather than presenting them as being separate individuals
when involved in separate activities. The frame of coherence that is used is
thus not merely a combination of the News pattern and the Human interest
pattern, but a pattern of its own, a new Gestalt.
The manner in which the Friendly news pattern has been presented so far
would indicate that the heading-plus-lead part and the golden-nugget part need
to be of different kinds: here, information vs. humour and other light aspects.
This is not, however, necessary. The International Herald Tribune (March 22-
23, 1997), for instance, makes use of the same discourse pattern, but stays
solidly within the realms of transmitting information. Here we first - starting
on p. 1 - get the issues mentioned: NATO, the G7, the Arms control debate
and START 3, and then at the end (p. 7), when all these have been reported on,
we get an informative golden nugget summary, where all these three are
connected into one - cf. text segment 5.
96 Jan-Ola Östman
... Implementation of
START-3, which would reduce strategic warhead stockpiles for both countries to
2,000 to 2,500, would be completed by the end of 2007.
Mr. Yeltsin seemed particularly concerned that he would be attacked at home for
having been bought off on NATO expansion with economic inducements that
include American backing of Russia's application to join the world Trade
Organization and membership in the Paris Club, a group of nations dealing with
debt and international credit.
Text 5.
8 Conclusion
In this study, I have attempted to approach coherence through socio-cognitive
understanding. I have furthermore argued that understanding on discourse level
can best be understood in relation to the concept discourse pattern. Part of
what we conceive of as the coherence of text/discourse is anchored in the kind
of global, cognitive, partly codified understanding we have of how to
categorize the text/discourse in question and how to hook this text/discourse
onto the cognitive frame of understanding that I call a discourse pattern.
Obviously, at this stage of the investigation, the viability of the notion
'discourse pattern' is a hypothesis; but I have shown that it is a plausible
hypothesis in need of further study.
Being socio-cognitive11, the discourse-pattern approach is neutral with
respect to speaker's/writer's intention and addressee/audience effect. But at the
same time, discourse coherence is dependent on the interplay of both. One of
the most important aspects that contributes to the effectiveness - and thus
coherence - of discourse/text is the extent to which an audience recognizes the
discourse/text as being of a particular type. The suggestion of this study is that
the notion 'being of a particular type' can be explicated in terms of discourse
patterns.
The suggestion that discourse patterns play an important role for
understanding and for coherence is in line with, and thus receives support
from, research indicating the importance of scripts and cognitive schemata, the
importance of formulas, the importance of means like discourse topics for
creating expectations in interpreting and construing discourse, and the
importance of what in dialogic terms would be that communication has to do
with being constantly engaged in 'projects' of focused understanding. The
feasibility of the notion discourse pattern can be further appreciated by noting
the large number of fields of applicability it has. For instance, the notion of
discourse patterning offers an additional level for detailed analysis of
intertextuality and recontextualization.12 Secondly, discourse patterns are akin
to 'grammatical constructions' - conglomerates of structural, semantic and
pragmatic information - on sentence level,13 and could from this point of view
be talked about as 'discourse construct(ion)s.' This perspective also enhances
the point that language is more formulaic than we might at first think - also on
discourse level. And thirdly, misunderstandings can profitably be investigated
from the point of view of what negotiative strategies interactants use when
their discourse patterns conflict and (temporarily) hinder understanding.
Analyses in terms of discourse patterns are, however, only one road to a
deeper understanding of discourse coherence. By taking this stance, I indirectly
indicate that coherence might be too large a notion to conceptualize as a whole
98 Jan-Ola Östman
Notes
1. In this study I am not concentrating on matters of local coherence - i.e. the connexity
between clauses.
2. The physical shape of texts - in the sense I talk about it here, as a 'pattern' - has naturally
received attention in literary studies, especially in poetics. In linguistics, however, the
shape of texts tends to be either dismissed, or taken for granted without further notice.
Some text-books do mention the importance of the shape of texts (cf. e.g., Bülow-Møller
1989: 51-52), but mainly in reference to, precisely, poetry.
However, in anthropological linguistic studies of indigenous languages and cultures,
shape has a more respectable status. For instance, Hymes (1996) suggests that the way
lines pattern in a narrative shows how human experience gets shaped differently cross-
culturally.
3. Werlich (1976) distinguishes between the narrative, instructive, argumentative,
descriptive, and expository text types. Other classifications include those by Longacre
(1976) and Kinneavy (1971). For an overview, see Östman and Virtanen (1995).
4. The Aesthetic Pattern needs to be dealt with in much more detail than is possible here, it is
the topic of a separate, forthcoming article.
5. Some others that I have found useful are, for instance, the In medias res pattern, and the
Legal document pattern.
6. One way to test the feasibility of the notion 'discourse pattern' as linked to cognition
would be to measure subjects' manner and easy of comprehension of a text when it is
presented directly after another text utilizing the same vs a different discourse pattern. The
hypothesis would naturally be that sameness in pattern enhances comprehension - but
sameness in text type or genre do not.
7. For instance, Cragg Hines's Houston Chronicle report (March 22, 1997: 1A, 16A) on the
Helsinki Summit was reprinted in the Santa Barbara News-Press (March 22, 1997: 1, 2)
with the last two paragraphs left out.
8. Similar distinctions are, of course, to be found in many other countries; the ones that first
come to mind - due to my personal bias - are Finland and Sweden,
9. At tins point I want to express my gratitude to Diana ben-Aaron for several lengthy
discussions that have greatly enhanced my understanding of news discourse.
10. Readers who would not normally admit that they read tabloids, can also in thns way read
'tabloidy' stories without changing newspaper. If my analysis is correct, this also means
Coherence and Discourse Patterns 99
that some readers - e.g. somebody who is not at all interested in US - Russian
relationships - might be inclined to jump straight to the end section of the news item, to
read the news report 'funnies' - the 'friendlies' if you like. It may be a coincident, but
sections (b) and (c) continue the news item, next to each other, up on the same page. Even
though this is speculative, we might suggest that these sections were moved up in order to
be easily accessible for readers heading straight for the lighter sections.
11. On the need for both embodied and situational cognition, see e.g. Östman (forthcoming).
12. For a recent overview of the importance of recontextualisation, see Linell (1996).
13. On Construction Grammar, see e.g. Fillmore, Kay and O'Connor (1988).
14. 'Prototype' here specifically refers to the fact that discourse patterns have not been set up,
nor related to each other in terms of necessary and sufficient conditions and similarities.
15. For a view of coherence as one of three parameters of pragmatics as implicit anchoring,
see Östman (1986). The relation between pragmatics and ideology is also touched upon in
Kuusisto and Östman (1998).
References
Auer, P. 1995. "Context and contextualization". In J. Verschueren, J.-O. Östman and
J. Blommaert (eds), Handbook of Pragmatics, 1995 Installment. Amsterdam:
Benjamins. (19pp.)
Biilow-Møller, A.M. 1989. The Textlinguistic Omnibus: A Survey of Methods for
Analysis. Copenhagen: Handelshøjskolens Forlag.
Fillmore, Ch.J., Kay, P. and O'Connor, M.C. 1988. "Regularity and idiomaticity in
grammatical constructions: the case of let alone". Language 64: 501-538.
Fries, U. 1990. "A contrastive analysis of German and English death notices". In J.
Fisiak (ed), Further Insights into Contrastive Linguistics. Amsterdam:
Benjamins, 539-560.
Galtung, J. 1979. Papers on Methodology. Copenhagen: Ejlers.
Goodwin, Ch. and Duranti, A. 1992. "Rethinking context: an introduction". In A.
Duranti and Ch. Goodwin, Rethinking Context. Cambridge: Cambridge
University Press, 1-42.
Hymes, Dell 1996. Ethnography, linguistics, narrative inequality: Toward an
understanding of voice. London & Bristol: PA: Taylor & Francis.
Kaplan, R. 1972. The Anatomy of Rhetoric. Philadelphia: Center for Curriculum
Development.
Kinneavy, J.L. 1971. A Theory of Discourse. Upper Saddle River: Prentice-Hall.
Kuusisto, P. and Östman, J.-O. 1998. "The media as mediator. How foreigner
discourse constructs ideology in Finnish newspapers". In J. Blommaert (ed),
Political Linguistics. Amsterdam: Benjamins.
Lakoff, G. 1987. Women, Fire, and Dangerous Things. Chicago: University of
Chicago Press.
Linell, P. 1996. Approaching Dialogue. Talk and Interaction in Dialogical
Perspectives. Linköping: University of Linköping.
Longacre, R.E. 1976. An Anatomy of Speech Notions. Lisse: Peter de Ridder.
100 Jan-Ola Östman
1 Introduction
This paper proposes a new term semiotic spanning and suggests that it might
be useful when we try to describe semiotically what is going on in
linguistically complex situations such as conferences. I have for a long time
been interested in academic writing in English and improving non-native
writing, especially by young scholars, and thus I have in this context been
interested in cohesion and coherence. Cohesion operates within texts and is
realised by cohesive links; coherence is a situationally determined relation,
linking up with the interpretative contexts of texts. I have also for a long time
been interested in how conference papers are written and presented and how
the discussion evolves after the paper. Here, too, the notions of cohesion and
coherence play a role - no participant at a conference wants to transmit an
image of her/himself as an 'incoherent' presenter/discussant. Yet, there are
times when we as conference participants wonder what the presenter's/
discussant's point really is and how it links to what has previously been said.
The notions of cohesion and coherence are helpful when showing individual
writers/presenters how to improve their texts. Yet for my present purposes
these notions seem to be too text and context-bound, i.e. too much tied to an
individual text and its relation to the immediate context of production.
This paper aims to demonstrate that we must extend our interests. We
should aim to go beyond textual cohesion and coherence and focus on how
texts are interrelated semiotically. Some might argue that we have already
developed such perspectives, as the notions of intertextuality and genre
suggest. But the kind of aspect that I would like to focus upon here is not, in
my view, covered by those terms. The notion of intertextuality, grossly
simplified, seems to cover the 'taking over' of certain features (mostly
ideational content) from one text to another, an appropriate example being an
102 Eija Ventola
editorial text that is based on one or two articles that have previously appeared
as news reports. The notion of genre seems to cover the relationship of texts of
the same kind as tokens of a certain class of texts. So, for instance, articles in a
linguistic journal may belong to the same genre (research articles) or agnate
genres (research articles vs. review articles). But what I am trying to present in
this paper is a more extensive view on how texts relate to each other by
spanning semiotically, i.e. linking up with various kinds of existing and
experienced texts (and other semiotic modalities) and creating new semiosis
through these links. This approach is necessarily dynamic as it spans across
various texts and situations. The closest thinkable comparison might be
'surfing in the Internet in the hypertext world of links' and creating something
new through the process, but 'new' not in an individualistic sense (as for
example a path created by the hypertext reader), but in a collective sense.
I shall approach semiotic spanning by looking at why the other above-
mentioned terms cohesion, coherence, intertextuality and genre are not
sufficient in explaining the kind of phenomena I have in mind. Sometimes the
spanning can, for example, be perceived between a conference paper and its
discussion, but it can be understood more extensively as well. Semiotic
spanning can be seen to function between various instances of genres within
the speech event - the different conference papers and discussions within one
section or during the whole of the conference when the discussions are built
upon previous papers and discussions. It is also part of the other kinds of
genres at conferences, e.g., when papers become part of other conference
genres - dinner-table talk, coffee-time chats - what we call talking shop. And of
course semiotic spanning exists between the presented paper and its source
materials as well as with the final written version of the paper. In this paper I
will focus on a number of issues: 1) semiotic spanning between a conference
talk and its sources, 2) the spanning between the talk and its discussion, 3) the
spanning between these texts and the texts of other talks, their sources and their
discussions and the conference as a whole, 4) the spanning between the
forthcoming written and spoken texts after the presented talks and their
discussions at a conference and the influence they in turn have on other papers
written for presentations and publication in the future.
(1)
1. Chair: Wenn nicht würde ich vorschlagen, daß wir fortfahren er mit dem
Vortrag von Frau XXX der diesen Sektor der wissenschaftlichen
Hinterlassenschaft in Form der Sammlungen... der Biographie also eine ganz
wichtige Sache behandelt
At the beginning of example (1), since no more questions are asked about
the previous paper, the chair hands the speaking turn over to the next presenter,
Frau XXX (a native speaker of American English), who after the initial
remarks about the language of the presentation starts producing what we in the
academic world would consider an appropriate beginning of a paper. Let us
first focus on its cohesion.
3. J Cohesion
and ask questions about the collections]. The question in (2) concerns a
subcategory of the Forster collections.
(2)
1. Discussant: I'd like to ask a question. You can rather ignore it probably. Can
people visit erm Göttingen the the collection the collection?
2. Presenter: Pardon?
3. Discussant: Can people visit the- these collections which are at Göttingen. I
mean, if I- Is there- And everyone can go there?
(3)
The next most important collection is the Institut und Sammlung für
Völkerkunde in Göttingen. This collection was sold to Göttingen in 1799, after
Reinhold's death. Its some 157 listed pieces is similar in scope and variety to the
collection in Oxford and was apparently Forster's own systematic collection that he
had retained for himself, although it does not include many unique objects and
diminished over the years by giving away and selling pieces.
3.2 Coherence
I shall again use Halliday and Hasan's views for discussing coherence (but I
also acknowledge that other scholars have explored coherence extensively
from its various aspects, cf. this volume). Coherence can very briefly be said to
be about the relationship the text has to its context of situation:
106 Eija Ventola
...texture involves more than the presence of semantic relations of the kind we refer
to as cohesive, the dependence of one element on another for its interpretation. It
involves also some degree of coherence in the actual meanings expressed: not only,
or even mainly, in the CONTENT, but in the TOTAL selection from the semantic
resources of the language, including the various interpersonal (social-expressive-
conative) components - the moods, modalities, intensities, and other forms of the
speaker's intrusion into the speech situation... A text is a passage of discourse
which is coherent in these two regards: it is coherent with respect to the context of
situation, and therefore consistent in register; and it is coherent with respect to
itself, and therefore cohesive.
(Halliday and Hasan 1976: 23)
In (1), we notice that when the presenter starts her presentation she
produces a speech act which we may interpret either as an apology (feeling
apologetic because she cannot present her paper in German) or as a joke. This
part of her presentation does not relate in any of the ways described above as
cohesion to the chair's turn or to her own paper, except for one lexical item
talk. Yet, no one with experience of international conferences considers such
speech acts as 'social niceties', jokes, etc. as odd and out of place. So,
contextual coherence in the communicative situation explains how the parts of
the presenter's text which lexicogrammatically do not seem to go together can
actually be seen to be linked. The same applies to the chair's turns, when he
introduces the speaker and opens and leads the discussion. His turns are
"coherent with respect to the context of situation, and therefore consistent in
register" (Halliday and Hasan 1976: 23), in terms of Field (relevant social
action), Tenor (relevant social and interactional roles) and Mode (channels of
communication, medium), and thus seem to form a unity that is contextually
coherent. Yet at times, especially during discussions, this contextual coherence
is put to test.
Achieving cohesion and coherence seems a complex process in the
context of conference papers and demands hard work from all of the
participants. When the presenter drifts off his/her topic or keeps jumping back
and forth topically, we may find it hard to build up links with what the
presenter is saying and the initial topic of the paper, and we may consider this a
failure in maintaining textual cohesion. But we may also experience breaches
in contextual coherence. Sometimes at conferences we find it hard to link up
the content of a paper with the global topic of the conference (if there is one),
with what the other speakers present in their papers and how their own views
come into play. As discussants we do not always see things in the same way
and even seem to have difficulties in understanding what the other discussants
want to say. In spite of the context of situation, coherence seems to be lost, as
in (4) where the presenter not only has difficulties in understanding what the
Coherence and Conference Language 107
discussant is saying, but also finds it difficult to establish common ground with
her. His linking differs from the discussant's.1
(4)
1. Chair: Ja bitte
'Chair: Yes please'
2. Discussant: Ich wollte nur... eine kurze Be Bemerkung. Sie haben von der
erm team teamship oder oder sagen wir nicht so erm oder die enge erm
Beziehung zwischen Vater und Sohn wissenschaftlich jetzt. Und meine Frage
wäre in wie fern man das auch so erm ... erweitern könnte also jetzt politisch
gemeint also ich meine, daß Georg Forster, der Sohn, dann eine völlig andere
erm einen völlig anderen Weg gegangen ist erm jetzt was seine politische
Entscheidung anbelangt. Also das heißt nicht daß der Vater da auch eine Rolle
spielen erm könnte aber ich glaube schon, daß erm solche biographische
Erklärung Erklärungen nicht ausreichen, um dann Forsters politische, jetzt
Georg Forsters politische Entwicklung zu erklären. Das ist wieder mein
Problem.
'Discussant: I just wanted a brief remark. You have {no main verb 'talked'}
about the erm team teamship or or should we not say about so or a close erm
relationship between father and son now scientifically. And now my question
would be in how far one could erm expand this, now in a political sense, that
is I mean that Georg Forster the son then went {'chose'} a totally different
erm a totally different way in erm what now concerns his political decision.
That does not, however, mean that the father would not be playing erm a role
there, too, but I think that erm such biographical explanation explanations are
there not sufficient to explain then Forster's political now Georg Forster's
political development. That again is my problem.'
3. Presenter: Have I understood you rightly that you think that Johann Reinhold
Forster had no influence on the political thinking of George
4. Discussant: No I meant perhaps he could have had
5. Presenter: *Oh I'm -
6. Discussant: *but not that much because when you erm read when you know
of erm Johann erm Reinhold's erm position after George Forster's erm erm
you said that in English erm
7. Presenter: Arrival back in Germany
8. Discussant: No I mean in Mainz erm
9. Presenter: Ya
10. Discussant: And after his death its erm its quite complicated and well... But I
think we can't discuss about this because it's erm we are not erm it's not
possible to discuss what was going on in their minds
11. Presenter: *No other than -
108 Eija Ventola
12. Discussant: *And I think that my problem is only how far if we consider this I
I think your task is very important I mean erm what concerning Johann
Reinhold Forster because he's been very erm erm
13. Presenter: *neglected
14. Discussant: *neglected ...**aha
15. Presenter: **and is still ya
16. Discussant: in England...where you you can still- well I'm Portuguese so I
can speak sort of erm
17. Presenter: Ya
18. Discussant: as an outsider
19. Presenter: Ya
20. Discussant: But I think if we insist on that point I think that we forget Georg
Forster as a revolutionary
21. Presenter: *Ya
22. Discussant: * That's my point. It's too much
23. Presenter: **Ya
24. Discussant: **scientist and
25. Presenter: Yes
26. Discussant: and zu wenig Revolutionär.
'Discussant: and too little of a revolutionary'
27. Presenter: Ya. I I I have to say quite frankly in the Southern hemisphere
that some of us here are interested in him primarily as a scientist. You can
keep your revolutionary activity if you like I mean we are interested but for us
erm and I'm speaking not speaking personally because I'm interested in the
whole of the life but here's the problem. ... They were thrown out of England,
their forefathers were thrown out of England for political reasons
{the asterisks and the underlinings indicate simultaneous speech}
4 Intertextuality
Intertextuality is mostly seen in terms of'sharing ideational content'. A typical
example of intertextuality in the context of conferences is perhaps seen in
terms of quotations and references to other authors' work in the papers. Parts
of texts become parts of other texts; the information is condensed, reformulated
and reshaped to fit the purposes of the author (cf. an editorial based on news
reports or the background and theoretical sections of academic papers).
According to de Beaugrande (1980: 14), this kind of intertextuality involves
"experientia knowledge of specific, actual occurrences" (his italics), and he
gives the following as examples of where we find intertextuality of this kind:
summaries, protocols, continuations, replies, parodies. The notion of
intertextuality has generally been used to describe the relationship between
written texts - texts that are products (for a detailed discussion, see e.g. de
Beaugrande and Dressier 1981: 188-214). Thus, looking at intertextuality at
conferences would mostly mean considering the relationships between the
presenters' written papers and those texts they have referred to for writing their
own texts. But this view is hardly sufficient since what is presented and
discussed at conferences not only relies on written texts, but on experiences
gathered through other semiotic means as well.
110 Eija Ventola
5 Genre
Martin and others have elaborated Halliday and Hasan's (1976) views on
textual unity in terms of cohesion and coherence and text construction and text
relatedness in terms of genre and register (see e.g. Eggins 1994; Martin 1992;
Ventola 1987). The benefit of the notion of genre in this context is that it could
offer us a dynamic view to generic unfolding of goal-oriented social activities,
thus allowing discourses which unfold slightly differently still be considered as
realisations of the same genre. It also allows different registerial realisations, at
least to a degree, during the unfolding process and guarantees appropriateness
in terms of field, tenor, and mode choices.
The choices made on the genre and register level have repercussions to
the ways texts are realised as discourse, lexicogrammatically and
phonologically (for examples and discussion, see Martin 1992). When we think
Coherence and Conference Language 111
- in other words, for explaining the various kind of linking and expanding of
semiotic information to other communicative contexts.
6 Semiotic spanning
What has been suggested above is that the previously discussed notions of
cohesion, coherence, intertextuality and genre are insufficient to explain in
terms of linking what goes on before, during and after conferences, what takes
place at conferences at all levels (including cafeteria talk about the papers) and
what keeps conferences together as social events. We do not sufficiently know
how such events as a whole contribute to 'the discourse world', and as linguists
we have never really attempted to describe all these links. The proposed term
semiotic spanning might be useful to explain this kind of complexity in the
world of discourses and the various combinations of semiosis that we
encounter. We should be trying to explain what makes conferencing
linguistically possible, how its influences are felt and developed afterwards and
what changes conferencing consequently initiates in the world of discourses.
We know very little about what it is that allows us to bring in so many aspects
into conference situations and yet (mostly) achieve some unity in knowledge
and understanding during and after the conference, and then take a step
forward in our understanding how the world (of discourses) functions.
Conferences as social events provide 'life' to conference papers and their
discussions (formal and informal). After the conferences, texts 'live on' in
form of other similar professional genres: in following discussions, in
conference volumes, articles, in the quotations in other people's articles,
reviews, in popularisations of papers, and so on. We even need to be able to
describe what is going on between related and non-related genres, i.e. how, for
example, 'scientific knowledge' presented in a paper at a conference gets to
become a part of a coffee/dinner-table discussion or a part of a newspaper
report or a school essay. This is where perhaps the notion of semiotic spanning
will prove to be useful.
It seems that so far we have hardly begun to adequately address all the
aspects of conferencing. Although some research has been done on the drafting
stage of writing papers, we still do not know enough about what conferencing
preparation entails - i.e. how the presenters combine the material they obtained
from different sources through different semiotic modalities (including
internet) with their own earlier reading and acquired knowledge in preparation
of their papers. Further, no systematic study has been conducted to see how the
activities of the presentation, giving handouts, showing transparencies or
slides, playing audiorecordings or videorecordings, etc. link with the
conference situation and the papers the participants present there, and how
Coherence and Conference Language 113
conferences generally use wordings created for other textual functions for their
own purposes and thus create a semiotic span between the original text and the
one created by them.
With this kind of spanning the presenters often bring 'knowledge/views'
established by others and incorporated with their own to the attention of the
audience. Sometimes, however, the presenters slowly build up the views/things
that they want to share with the audience. Thus, we notice for instance in (1),
that for the chair (and perhaps for many others in the audience) 'the Forster
collections' seem already to be 'shared knowledge' (as indicated by the
wording when introducing Frau XXX). But the presenter cannot rely on this
kind of semiotic spanning functioning for all of the audience. She must build a
'shared view' with his/her listeners, and we see it happening in (1). The
presenter establishes technical terms and gets them accepted as 'referable
things', which then can be taken as topics for discussion by the audience. In
this way the process contributes to the semiotic spanning between the paper
and its discussion. If we look at Frau XXX's presentation as a whole, the most
important items discussed in the whole paper and in the discussion are the
Forster Collections - the artefacts that the Forsters collected during the second
Cook voyage around the world. They have been distributed to various places as
separate collections: Oxford, Göttingen, Görlitz, Mittau and Stockholm.
The collections can be related back to the transitivity structure that goes
back to the events in the Forsters' time, namely the fact that the Forsters
collected artefacts, which implies the following congruent transitivity
structure: Active Material Process ˆ Goal3. At the beginning of Frau XXX's
talk, this structure is also prominent, although grammatically it does not always
get realised congruently as a material clause (and as we have already seen the
chair uses the nominalised form in his introduction of the presenter). Example
(5) shows the first mentions of the activity of 'collecting' in Frau XXX's
paper.
(5)
As far as the Forsters were concerned their purpose on the second voyage was to
contribute to the scientific knowledge of the world through collecting and
publication. Strictly speaking, there were problems here also. Many of the new
plants and animals had been previously discovered and collected during the first
voyage by Banks and Solander. Most significantly the Forsters' biological and
anthropological specimens, although collected systematically, were widely
scattered and there was no systematic publication of the botany, zoology, geology,
or anthropology by the Forsters or anyone else... (later in the text). Thus our focus
on the present day is the distribution of the ethnographic collections made by
George and Reinhold Forster and the knowledge that can be gained from the
objects and the writings about them.
Coherence and Conference Language 115
When one examines the final stage of (5) and examples (2) and (3), it is
obvious that the material process has been established as a referable Thing. The
activity sequence has become a technical term in Frau XXX's presentation and
is, for example, realised in its nominalised form as the Forster Collection, i.e.
Deictic the Classified Forster Thing Collection or as the largest, widest
ranging and most important collection made by the Forsters (Deictic (Epithet
Epithet Epithet) Thing Qualifier). This Thing is further subclassified
into different kinds of collections, named after the places where the items are
kept today (e.g. in Göttingen in example (2)).
By the time the discussion starts, this technical term and all the other
ones similarly created, have been accepted by the audience as things that one
can refer to, talk about, argue about. For instance, in (2), when the discussion
starts one of the conference participants asks Frau XXX: Can people visit eh
Göttingen the collections?. The created classification becomes the focus in the
discussion. This kind of 'establishing shared view' (mostly through what
Martin 1992 called ideation) and then referring to it (Martin's identification)
helps us as conference participants to think that we are talking about the same
things and that we understand each other and the argumentation. It establishes
links between the paper and its discussion and the other papers and their
discussions presented afterwards. In our example text, other researchers can
refer back to the collections mentioned by Frau XXX in their papers and thus
create a new node in semiotic spanning that helps to unify the conference as an
event.
Frequently, however, discussants do not just establish semiotic spans
about shared information, as in (2), but rather make links with their 'semiotic
worlds', as in (4), where the discussant tries to bring forth the political aspects
of Georg Forster into the discussion, but this spanning is rejected by the
presenter. Here then we have a confrontation of two semiotic spannings - both
the discussant and the presenter link to the semiotic worlds which they have
experienced and do not seem to be able to reconcile their views.
In the discussion following Frau XXX's paper we indeed have several
instances where semiotic linking is made to some other things than the shared
'world' that she establishes in her presentation. Other expertise is brought into
light in (6).
(6)
1. Chair: {after a fairly long turn by the chair himself and the silence, he gives
the turn to the discussant} ... (7 secs) Ja?
'Chair: Yes'
116 Eija Ventola
2. Discussant: Gestatten Sie mir einige Bemerkungen zu dem Besuch auf der
Osterinsel... erm die beiden Forsters wie de- wie der größte Teil der ganzen
Besatzung waren sehr krank. Erlauben Sie mir, daß ich das sage. Ich bin
nämlich Mediziner. Sie hatten alle Skorbut fast alle und der alte Forster
beschreibt ja wie mühselig er gehen konnte. Das lag an den dicken Beinen, die
er hatte...Ödeme, die sich da gebildet hatten in Folge dieser C-A A-
Vitaminose, das war's übrigens nicht allein. Die Nahrung war qualitativ und
quantitativ unzureichend. ... Deswegen ... ist ja auch Kapitän Cook so rasch
wieder abgereist... weil sie nicht genug zu essen bekommen haben ... und der
... jüngere Forster Georg ... der war ja damals noch gar nicht so alt noch. Der
Junge war ja sehr jung ... erm der eine war ja erst 18 Jahre ... da ... erm der
jüngere war so krank, daß er also kaum sich an Land schleppen konnte.
Vielleicht ist es doch nicht ganz uninteressant zu sagen wie diese Fülle von
Materialien, die die beiden schwerkranken Leute in zwei Tagen gesammelt
haben. Das finde ich doch immerhin beachtlich.
'Discussant: Allow me {to make} a few remarks concerning the visit to the
Easter Island ... erm both Forsters as well as the- as well as the main part of
the whole crew were very sick. Allow me to say that. I am namely a doctor.
They all suffered from scurvy, nearly all and the elder Forster describes how
difficult it was for him to walk. That was due to the swollen legs he had.
oedema that had developed for all as a result of this C-A A-Vitamin
deficiency, that wasn't by the way the only reason. The nutrition was
qualitatively and quantitatively insufficient... Therefore ... Captain Cook
actually left so quickly again ... because they did not get enough to eat... and
the ... younger Forster, Georg...who at that time was not that old yet. The boy
was very young ... erm he was just 18 years ... when ... erm the younger was
so sick that he could hardly drag himself on the shore. Perhaps it is not too
uninteresting to mention what a wealth of material these both of these two
seriously ill people collected in two days. That I still find all the same
remarkable.'
In the next example, (7), the coherence and the semiotic spanning is
further complicated by the fact that the discussant's remark is not at all
addressed to the presenter, Frau XXX, but to the chair who had presented his
paper before her at the conference. Thus, to see the discussant's turn it in the
right light, the audience (and we as the readers of the transcript) have to
establish links with the chair's paper and his semiotic world as well as the
discussant's world.
(7)
1. Chair: Ja noch weitere Fragen (7secs)
'Chair: Yes, some further questions'
'Discussant: Not a questions but an addition to you Mr. X (=the chair) to your
very valuable presentation...I would like to say that what you have worked out
that the importance of the statements and erm notes {written down} by both of
the Forsters on the Easter Island ...and how you interpreted culture and put it
into the ( culture?) before. The same applies also to the Gesellschaftsinseln
and also applies as a an example of ...and I would then like to add briefly to
the example that you yourself gave. You said that with (ee's ?) ... that in the
18th century there perhaps was a very strong connection with magic We have
this religiousness of the people still today on the islands, when I visited the
islands nineteen-seventy-two ... seventy-three and eighty-five during my
research trips, following the patterns of the belief of overall (?) spirit of the
death ... still existing in connection with the kind of religiousness of the
people that has been kept until today. This is what Forster referred to in the
year 1774. Thank you.'
We must ask ourselves to what extent the discussants can rely on the
conference participants' willingness to see their remarks, questions, etc. as
118 Eija Ventola
relevant and coherent and whether the audience can establish the appropriate
links to understand the point, especially when we are moving from discussing
one paper to another without much warning and prearrangement. Examining
semiotic spanning thus involves looking at how we link with other sources,
with other experiences and their expressions, and still keep what we say as
relevant and focused in terms of the presented paper and the point of the
discussion.
Discussions are examples of dynamic unfolding of discourse. They are
problematic in relation to their papers at least in two ways. Firstly, they cannot
occur independently as texts. If there is no preceding text, the paper, there
cannot be any discussion of it. Consequently, the discussion should always in
some respects be related to the paper and to the whole context of situation;
otherwise it cannot be perceived to be coherent. I have demonstrated how, in
the discussion, links are made far beyond the boundaries - to participants'
semiotic worlds. Often participants also 'prevent' successful semiotic spanning
with their actions. Questions do not get relevant answers, or the presenter
employs 'avoidance techniques', as s/he does not think him/her to be an expert
who can do the semiotic spanning in certain areas of knowledge.
Secondly, discussions have a capacity to develop far from the original
subject matter of the paper. The discussants simply perform long 'self-
presentations' that ideationally may have little to do with what has been said in
the paper - or at least the other participants do not seem be able to follow the
intended semiotic linking. Such comments may lead the discussion off the
intended focus and, consequently, the 'connection' with the presenter's original
messages is lost. Sometimes the discussion seems to 'side-track' dynamically
along its own paths and begins forming its own semiotics independent from the
paper, as almost happens in (6-7). Discussants start making other links to the
previous papers and their discussions, to the texts that conference participants
have read, to the other goings-on at the conference or in the outside world and
so on. It is then naturally the fonction of the chair to bring the discussion back
to its original course. Sometimes the questions asked offer the presenter the
possibility to side-track as well, for instance to present new material. This
happens in Frau XXX's case when one of the discussant asks her whether she
could see herself publishing a complete list of the items in the Forster
collections some time in the future, to which she replies: I actually- in my
paper in the section I didn't read to you is- I have a chart erm where the
collections are like islands they come from and in fact what objects what object
types there are and so....
So far semiotic linking has largely been exemplified in terms of evidence
of linking as seen through the linguistic realisations in the paper and its
discussion. The last example, (8), displays a link made evident to the audience
Coherence and Conference Language 119
through the spoken as well as through the visual mode. Frau XXX shows slides
to the audience twice during her talk. Frau XXX has visited the museums
where the Forsters' collections are located. Her experience of what she has
seen is captured in her slides, and here she is trying to share that semiotics with
her listeners. Example (8) displays part of the first SLIDE SHOW.
(8)
So I begin with- to show you some slides of the Pacific collections from the
Forsters and erm especially the collection in the Pitt Rivers Museum ... (27 sec
pause) can you hear me if I stand over here?...I start with Wörlitz. This is the
country house in Wörlitz if any of you have not been there...and these are the little
buildings in which the objects from erm the Forsters who have their own special
little building for the objects made for them and if as I said it is a really excellent
collection and when I visited it was still in quite good condition...This is the Easter
Island hand that was collected. It is a unique piece and now in the British Museum
but was given by the Forsters to erm the British Museum ... it's okay? This is the
Pitt Rivers Museum which is erm as you can see a rather old-fashioned museum
which is arranged by artefact type rather than by area ... And this is the catalogue
of curiosities that was written apparently by George Forster that accompanied the
collection when it went erm to the Asmolean Museum in the 18th century. Some of
the important objects can al- you can only figure out what they were used for by
looking at illustrations. This is an illustration by Hodges and note the little human
images on the top... of the erm I'm sorry ... the er what I wanted you to look at was
the- the- erm the man wearing this wonderful feathered head-dress and the one of
these unique pieces was collected by the Forsters and is now in the Pitt Rivers
Museum ... (7 sec) now I can't move it forward ... and look at the erm human
images and the front of the... erm erm the canoe and here are the two that were
collected by the Forsters both of them given erm to the Pitt Rivers Museum and
these were the only two human images that were collected apparently during the
second voyage.
discourse mode, which then puts different linguistic systems at stake in the
production. Frau XXX uses language differently in this part of her talk because
both visual and spoken discourse modes are active. This has several linguistic
consequences. Firstly, the realisations from the IDENTIFICATION systems at
the level of discourse are exophoric rather than endophoric4. The presenter
refers to the entities with exophoric pronouns; the reference relation has to be
interpreted contextually, e.g. this is the country house in Wörlitz. Secondly,
although the experiential content (the artefacts collected by the Forsters) does
not vary, the IDEATION is now also realised differently lexicogrammatically.
The prominent transitivity structure is a relational, identifying process with the
exophoric pronoun being the Token and the nominal group after the relational
process being the Value (see Halliday 1994), as in: this (Token) is (Relational,
Identifying Process) the country house in Wörlitz (Value). Furthermore, in the
slide show the logical function gets realised largely by additive
CONJUNCTIONS, as in: and (additional conjunction) these are the little
buildings... (Note in contrast that Frau XXX's prepared manuscript includes
various other types of conjunctions and metaphorical realisations of logical
relations.) Finally, the interpersonal aspect becomes more prominent in the
slide show - more choices from the NEGOTIATION systems are made and
realised lexicogrammatically: can you hear me if I stand over here
(MODALITY: MODULATION; mental process; Question; polar
interrogative); and note the little human images on the top of the erm... (mental
process; Command; imperative).
The factors of cohesion, coherence, intertextuality, and genre do not
explain, nor were they meant to explain, the ways in which Frau XXX
incorporates the slides which were taken under entirely different conditions in
which originally formed a different kind of semiotics. Through asking
questions about semiotic spanning we may find out what she had hoped to
achieve through this kind of linking and to what degree she was successful
Obviously she felt that this kind of realisation of linking could bring the
audience closer to the 'semiotic world' she had experienced throughout her
research. Yet, in the written paper she is forced to leave this kind of linking
out, and the reader of her article simply gets information about the collected
items in the appendix of her paper, where the items are listed according to the
museums where they can be found. It is certain that the interrelatedness of this
kind needs more detailed exploration, especially when we consider what lies
behind the 'preparing a paper for a conference' and what happens after the
conference when the conference paper is (re)written into a publication, when it
takes on a life of its own by being discussed, reviewed and quoted, and so on.
Coherence and Conference Language 121
7 Conclusion
When writing papers, we build various links with other discourses - i.e. the
sources, etc. of the papers. The discussions of papers at conferences then link
with the presented paper or other papers preceding it as well as the discussants'
own experiences of discourses (or other realisations of their semiotic world); a
cumulative increase of perspectives of knowledge and understanding is
achieved at conferences. How 'conference presentation texts and their
discussions' come about seems to involve a complexity of phenomena: reading
of other texts, viewing visuals, making experiments, questionnaires, films,
taking pictures and slides, etc. The process requires a new approach for its
study, and thus semiotic spanning has been suggested as a term that could
cover all of the aspects of linking I have addressed. Linking in conference
situations clearly takes place largely through linguistic realisations, although
frequently visual mode combines with the spoken or the written mode (as in
transparencies or slide shows). However, we also need to see how the papers
exist after their instantiation, how they participate in the semiosis of other texts
of different kinds of genres. We must start to research how different recipients
experience the same information presented in a paper: what is obvious to one
conference participant about the presenter's paper can to another person be
incoherent and devoid of semiotic linking with his/her semiotic world. The
function of a conference is to establish and build common ground. In my view,
as linguists we still have a fairly limited view on how we succeed or fail in
doing this. Extensive research has been carried out on both cohesion and
coherence in individual texts and/or similarities/dissimilarities of texts as
classes of texts. However, very little research has focused on how texts are
actually related, not in the sense of formal or generic relationship, but in the
sense of how they create a social event as a whole and how they build the
social event as a chain of events over time (e.g. how a paper is written over a
period of time and how it lives on later).
One might say we are dealing with something like an internet hypertext:
the way in which, at a conference, someone else has previously conducted the
research and built up the text resembles the text readers/users create when they
do the linking in internet. The presentation of the paper at a conference is
similar to that of an internet-user getting the first text on the screen. As soon as
the text starts unfolding, the conference participants start making their own
links. They relate what they hear to their own individual 'semiotics', and where
one establishes links, another may see no reason to establish any links at all
(and thus switches off). This process of semiotic spanning influences the flow
of the discussion: what questions are eventually asked and what comments are
made. It may also influence the next presenters, because they, too, have been
122 Eija Ventola
building up links in their thinking with what they have heard. Consequently
their 'prepared' text may go through radical changes during the oral
presentation. This kind of semiotic spanning on the part of the discussants and
the presenters actually results in a very interesting and innovating conference,
which ultimately motivates the participants to carry on their spanning on their
own. On its own each paper may not have a dramatic effect on the conference
participant, but what makes the conference valuable is the effect of the
interplay between all of the papers and their semiotic spanning within and
outside the social event of a conference. This spanning is cumulative as the
conference proceeds from one paper to the next, but due to the dynamic nature
of unfolding of papers and discussions, this aspect of conferencing has, to my
knowledge, not been researched at all.
Frequently we find that the sense of semiotic spanning created during a
conference during the papers and their discussions (formal and informal) is not
of lasting nature. Actually most conferences merely create an illusion of
temporary understanding and unity of an academic discourse community. Even
immediately after the conference the links between discourses by the members
of the community are reformed and changed. The conference situation creates
a momentary feeling of intellectual companionship and sense of common
understanding and experience, but unfortunately its effects may, remain very
short-lived and local. Yet, one could perhaps enhance the effects by closer co
operation in the publication and planning of future events around the theme of
the conference. When papers are published their effect should be cumulative;
in other words, they should influence the thinking of the conference
participants and other scholars and help them to understand the phenomena (of
whatever kind) around us. But ensuring that effect is practically impossible in
the publishing world.5
What remains to be said is that, when viewed semiotically, conferencing
remains an unexplored territory. Although not entirely defined and exemplified
in this paper, the term semiotic spanning has been suggested in order to allow
linguists and researchers to explore different kinds of relationships between
different social activities, activities which are sometimes not closely related,
yet influence and enhance our thinking in other fields.
Notes
1. The asterisks and the underlinings indicate simultaneous speech.
2. For a detailed discussion, see e.g. Ventola 1987; Martin 1992, especially Section 7.3.
3. For theory and analyses of transitivity structures and other kinds of verbal processes, see
Halliday 1994.
4. See Martin 1992 for a reference to these systems.
Coherence and Conference Language 123
5. A good conference publication should also perhaps resemble a hypertext in the sense that
there should be enough textual and semiotic spanning between the texts. Too frequently
conference publications are not built upon semiotic spanning and, consequently, the
publishers frequently are reluctant to publish conference volumes. They do not tend to sell
well because, out of the many articles, the potential buyers may only be interested in one
or two articles; i.e. as a whole, the conference volume does not form a 'semiotic whole'.
Today it is simply too easy to photocopy the few articles that one is interested in the
conference volume. One does not need to buy the whole book (or one gets the articles
from the internet). To create a publication that enhances 'semiotic spanning' is very
demanding for the authors and the editor. What the editor can naturally do is to put the
articles as they come in on his/her homepage and then ask the authors to rewrite their
papers by establishing necessary links with the other papers. Ultimately, then, the readers
will make their own connections when they read the final product.
References
Beaugrande, R. de 1980. Text, Discourse and Process. London: Longman.
Beaugrande, R. de and Dressier, W. 1981. Einführung in die Textlinguistik. Tübingen:
Niemeyer.
Eggins, S. 1994. An Introduction to Systemic Functional Linguistics. London: Pinter.
Halliday, M.A.K. 2 1994. An Introduction to Functional Grammar. London: Arnold.
Halliday, M.A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Martin, J.R. 1992. English Text: System and Structure. Amsterdam: Benjamins.
Ventola, E. 1987. The Structure of Social Interaction. London: Pinter.
Coherent Keying in Conversational Humour:
Contextualising Joint Fictionalisation
Helga Kotthoff
Universität Konstanz
1 Introduction
humorous discourse because they are more creative than in serious discourse,
both in production and reception. Humorous utterances are often not what is
contextually expected; violating normative expectations is indeed an essential
characteristic of humour. In conversational humour we quickly find extremely
complex inferential paths used to create "sense in nonsense". In explaining
what happens in humorous discourse, we must employ conversational
inferencing in the sense of Gumperz (1982) (i.e., 153: "the situated and
context-bound process of interpretation, by means of which participants in an
exchange assess others' intentions, and on which they base their responses")
and Cook-Gumperz and Gumperz (1994), based on contextualisation
procedures.
The keying of an activity as "humorous" is largely accomplished by the
use of marked prosodic cues, in conjunction with laughter, rhetorical
procedures of surprise organisation, pragmatic incongruities, shifts in
politeness standards (Kotthoff 1996a), and a general decrease in truth-value.
Through the above-named processes, speakers index a play-space: Bateson
(1954/1972) has shown that even primates can distinguish between a serious
bite and a playful one. Play framings are highly significant in ascribing a
humorous keying to an event. Keyings influence the reception of an activity;
they are closely related to affect management and the manipulation of truth
values.
Many theories of discourse coherence work with the notion of relevance
(e.g., Grice 1975; Dascal 1977; Sperber and Wilson 1985; Jucker 1993).
Discourse is coherent if relevance can be constructed for its parts. Sperber and
Wilson (1985) define relevance in terms of contextual effects and processing
effort. Contextual effects are achieved when newly-presented information
interacts with a context of existing assumptions. In humour, newly-presented
information often does not interact with the context of existing assumptions;
otherwise it would not be surprising (Mulkay 1988; Kotthoff 1998).
Furthermore, Sperber and Wilson theorise that the greater the contextual
effects, the greater the relevance, and also the smaller the processing effort, the
greater the relevance. Hearers pay attention only to information that enriches
their set of assumptions at low cost.6 However, we find that in humour listeners
must make an extra processing effort, which means that their costs are not low.
Why then are they willing to pay these extra costs?
It is necessary to distinguish between jokes which depend on punchlines
(German Witzigkeit) and comical discourse (German Komik). Only punchline
joking permits written representation. Because it is strongly based on lexico-
semantics, its humour potential can be inferred within the text. Everyday oral
humour often combines punchlines and comical keying. Punchline-joking is
based on lexico-semantically induced cognitive inferencing, the comical works
128 Helga Kotthoff
with keyings which play down the seriousness of utterances and modify their
reception, thereby influencing cognition. When speakers cannot rely on lexico-
semantically steered cognition alone, the humorous must be contextualised on
the level of performance. Contextualisation focuses on the performance7 aspect
of humour. Our attention, as researchers, thus shifts from the structural analysis
of bounded texts and lexical semantics to situated discursive practice. In
informal humour discourse, special contextualisation procedures lead to special
cognitive inferencing through which coherence is constructed for the
humorous. Comical discourse alludes much more to exclusive group
knowledge, as readers of transcripts can testify. Often they can only understand
humour potentials after being given background information by the researcher.
Both forms of humour demand extra processing efforts. Punchline-joking
works with the surprising bisociation (Koestler 1964) of frames or scripts;8
comicality works with double perspectivation (more on this below).
Let us look at one sort of effort typical of humour based on creating
punchlines.
He: Your nagging just goes in one ear and out the other.
She: That is because there is nothing in between to stop it.
for poetic effects. The same can be said for humorous effects and might be
integrated under the rubric of aesthetic pleasure.
There is agreement among coherence researchers that we accept extra
efforts if we can thereby expect to gain extra effects such as joy, aesthetic
pleasure or enlightenment. But in the literature this is chiefly reserved for the
'higher' arts.9 Thus Graesser, Person and Johnston (1996) consistently limit
aesthetic text dimensions to canonical literature. They cite numerous authors
who support the following thesis: "According to the polyvalence assumption, a
reader constructs multiple interpretations of a sentence or excerpt in a literary
text, whereas there is only a single bona fide interpretation in a nonliterary
text" (1996: 9). My article questions this delimitation. Or expressed differently:
Everyday discourse cannot be equated with nonliterary texts. Humour is a form
of oral art integrating many aesthetic strategies.
Like supporters of the contextualisation theory, Sperber and Wilson
(1985: 137-140) disagree with the traditional view of context as given prior to
the act of communication. They assume that processing new information
involves the selection of an adequate set of background assumptions which
constitute the context. They do not, however, sufficiently clarify the manner in
which a frame of interpretation is jointly set up for the rest of an utterance's
linguistic content. Taking conversational processes of contextualisation into
consideration is a way to enrich the cognitivist approach by clarifying the way
inferencing is steered in talk-in-interaction. Sperber and Wilson's approach to
communication and cognition is similar to Grice's (1975), in that they
underestimate the surface form of utterances. Humorous discourse illustrates
how implicatures and inferences depend on the concrete surface form of
utterances.
Conversation analysts have made valuable contributions to the analysis
of the concrete surface of speech activities. They focus on such questions as
when precisely people laugh in conversation, how this changes discourse and
to what extent laughter represents an orderly conversational activity
(Schenkein 1972; Sacks 1974; Jefferson 1979, 1985).
I would like to show the peculiarities of humorous coherence in a genre
which is not categorised in everyday discourse. We speak of such activities
simply as joking; and yet they display patterns which can be uncovered from a
scientific etic perspective.10 It is central for the coherence of the genre of "joint
fictionalisation" to extend the initiated humorous potential, to extract more and
more new comical aspects from the preceding talk. It has been variously
argued that there is not just one principle of coherence formation, but several
(Giora 1997). My work supports this claim. The coherence of keying and the
co-construction of the comical are special dimensions of oral humour.
130 Helga Kotthoff
3 Discussion of Data 1
Let us examine a conversational episode which contains joint humorous
fictionalisations. The episode is taken from a corpus of twenty dinner
conversations among friends taped in Austria, Germany and Switzerland in
1994-1995. The data is analysed using conversation analytic techniques
enriched by ethnographic methods (Kotthoff 1998).11 Some of the joking
activities have recognisable patterns, others do not. Overt joint fictionalisation
is regarded here as a joking genre if comical perspectives or punchlines are
created in its course.1 But there are naturally also covert and pre-patterned
fictionalisations in other forms of joking. Every standard joke is fictional;
already in the initial stage of telling a joke it becomes clear that the
interlocutors are about to enter into a closed fictional world with its own
coherence rules (Sacks 1974). Joint fictionalisations, however, leave the realm
of reality step by step, one fictionalisation building the ground for next, thereby
enhancing comicality.
The following episode was taken from a conversation recorded in a post
graduate student group in Vienna. In conversational joking we must always
share the background knowledge of the interlocutors in order to understand the
humorous potentials (this is part of coherence work). The group is chatting
about Hermes Phettberg. 13 At the time the recordings were made, (1995),
Hermes Phettberg was a very popular television moderator in Austria and
Germany. 14 His popularity was based partly on the fact that he was unusually
corpulent, especially for a media personality; he was also a confessed
homosexual masochist and appealed to a more intellectual public.
Conrad (C), Hugo (H), Lilo (L), Peter (P), Renate (R), two or more persons (m)
Before the start of the transcript, someone had commented that due to his
excessive weight, Hermes Phettberg would never live to be seventy. He was
thus selected as a topic, but not necessarily for a humorous modality. Conrad
agrees in lines 1/2 and notes that no one knows how much Phettberg weighs.
Renate, with a slight question intonation, makes a concrete guess. Peter in line
4 offers an even higher estimate. So far the talk is serious. In line 7 Conrad
makes a statement about himself which demands extra coherence work to fit it
into the topical context. The last word contains a laughter particle. Some
participants react with laughter (8, 9), which suggests that they have processed
Conrad's statement as comical. Coherence is formed by them, although a bona
fide understanding of line 7 would not make sense. Conrad is referring to a
youth magazine, Bravo, which publishes so-called Steckbriefe (which in
English could be called "celebrity trading cards") containing all sorts of
information about film, pop and rock stars and celebrities. Here we have an
unusual combination of elements from the real life world. Phettberg does not
fit at all in the sensationalistic Bravo, which does not deal with intellectual and
unconventional personalities such as Phettberg. The interactants can thus create
a perspective in which it is funny to imagine Phettberg in Bravo. Line 7 does
not create a classical punchline, however. There is no sudden frame bisociation
with the typical implicit possibility to construct meaning in a frame that has not
been developed yet (Koestler 1964; Norrick 1993; Attardo 1994). Combining
Phettberg and Bravo sheds an unusual light on both, as will be discussed
below. Peter does not understand something (perhaps he could not process the
comical dimension), Hugo's question in line 11 is directed at Peter, who in line
4 made an assertion about Phettberg's weight. In the following Peter softens
his assertion somewhat. In line 15 Conrad agrees with him. Up to here some
Coherence and Conversational Humour 133
interactants have reacted in a way that vaguely suggests they have been
constructing a sense for line 7. Other speakers have elaborated the subtopic of
Phettberg's weight. In multiparty talk it is not essential that everybody
understands everything at all times. We see that Conrad's remark in line 7 is
greeted with laughter and is then otherwise passed over.
In lines 16/17 Peter returns with fantasising to the topical area of Bravo.
He draws on a topical potential which Conrad introduced in line 7. In lines
16/17 we witness special keying procedures. The phrase des wär was/that
would be something fonctions as an introduction to something which is marked
as unusual. The turn is syntactically and prosodically subdivided into three
phrase units which all have the same rhythm and intonation. The accent is on
the first syllable in each phrase; the intonation falls at the end of each phrase.
Rhythm and intonation shifts can indicate a new context.16 Here they go
together with semantic comicality; in this way a humorous keying is set up
which is not totally new in the context, but is intensified. The youth magazine
is known for celebrity cutouts (Starschnitt), in each edition sections of a picture
of a celebrity are printed, and readers can gradually piece together life-sized
star portraits like a puzzle. This sort of world knowledge is activated here.
Extended laughter by several persons in line 18 shows that something
considered funny has been processed.
Let us look at some details of formulation. In lines 16/17 Bravo is
introduced in headline style. This means that elements of Bravo (a journai has
headlines) are used not only to denote this magazine semantically, but also
simultaneously to stylistically evoke it. The syntactic and prosodic form itself
creates the semantic content via iconicity. This "likeness on several levels"
(Jakobson 1960: 369) characterises aestheticised speech.
Furthermore, the continuation of a very elliptical speech style is striking.
We talk about ellipses if at least one part of a phrase is omitted that can be
completed contextually. Ellipses usually contain the rheme, the new
information (Schwitalla 1997). In line 20, for example, the suggestion to put
Phettberg in Playgirl, a magazine takeoff on Playboy which features photos of
nude or scantily-clad males, as well as the evaluation no viel besser/much
better, are quasi brief spotlights on an already-created stage; lines 21 and 23
also cohere in form and content with lines 16/17.
The presented fictionalisation is absurd, since in the real world Phettberg
absolutely would not come into question as a teenage idol to be popularised in
Bravo. In line 18 the women in the group laugh. Lilo comments on the fantasy
that in the case of Phettberg it would take an especially long time to collect all
the parts (a Jahr/a year); she is alluding to his enormous girth. Through their
demand for active coherence formation, allusions (Wilss 1989) contribute to
further aestheticising the discourse. The active reception here shows that the
134 Helga Kotthoff
comicality was not only understood but built upon. In conversational humour
recipients directly participate in further shaping the talk-in-interaction as
humorous. Humour and fictionalisation thus become an ongoing joint
achievement. In line 20 Hugo further intensifies the fiction about Phettberg in
Bravo in the topical direction of Playgirl. Previously created images are now
rounded out with specific details. Peter would like to see Phettberg presented a
year long in this magazine as well. Several persons laugh (22). In line 23 Lilo
enlarges the life-size poster to one twenty-five meters high and laughs at this
absurdity.
Starting in line 24 Peter takes up a different aspect of Phettberg's girth
and erotic self-display. The star's vast, sagging belly conceals his genitalia
from view. Conrad and Hugo affirm this image. Phettberg and 'minimum age
restrictions' are a new combination which is now developed by other
interlocutors. Conrad in line 32 alludes to a children's pastimes: Malen nach
Zahlen/painting by numbers11 Peter formulates a further fictionalisation from
children's pastimes {das Phettberg Puzzle). Lines 32 and 33 again display
headline-style. The games are presented like an ad. Lilo laughs and utters an
interjection of disgust. Conrad, in pointing out that it is a
Lebensaufgabe/lifetime job, once more alludes to Phettberg's enormous girth.
All the fictionalisations draw on a common knowledge of entertainment
media, thereby making coherence easy. The interactants successively intensify
the absurdity. The fictionalisations have a meta-message: Hermes Phettberg,
who with his "Nette Leit-Show'V'Nice People's Show" 18 carries on an
entertaining, witty type of media conversation, is relentlessly marketed in the
fantasies of the young Viennese. His body shape is the pivot point for
numerous ideas. The interlocutors do something Phettberg himself often does;
but they do it so-to-speak in a diametrically opposite manner. In numerous
interviews, Phettberg has referred over and over to his body, body feel and
sexuality. He achieved his popularity to a considerable degree by staging
himself as an anti-type. Phettberg contradicts several norms of the yellow press
world. He notoriously presents himself in interviews as overweight, unkempt,
homosexual and masochistic - thus aiming at a shock effect which appeals to
certain segments of the intellectual public. It is amusing to imagine him
integrated into the yellow press world as though he were a typical TV star or
celebrity. Thereby the Viennese interlocutors also communicate that they find
Phettberg's self presentation inconsistent and contradictory. Consequently,
distance can be simultaneously displayed toward both Phettberg and the yellow
press. The participants indirectly show both their knowledge of media contents
and also their critical attitude to them without explicitly making an evaluation.
By means of joking the evaluation is jointly performed rather than denoted.
Here the norms of the magazine world are violated and, at the same time, social
Coherence and Conversational Humour 135
4 Discussion of Data 2
The next episode of humorous joint fictionalisation was taped in the same
Viennese group. Lilo begins to tell a story from her childhood in a marked way
with interspersed laughter particles. Laughingly she describes some difficulties
her parents had in feeding her. These are removed from the domain of the
realistic through picturesque exaggeration {einmal im Monat t Küche
ausgwaschen/once a month scrubbed the kitchen from top to bottom). Someone
in the group immediately reacts with brief laughter.
Conrad (C), Elisa (E), Gerda (G), Martin (M), two or more persons (m), Lilo (L),
Peter (P), Sabine (S)
He formally presents details of the scene of throwing food at the wall. His use
of alliterations is striking (weiße Wand, roter Rübensaft). Line 37 is again
constructed speech. With a remarkable register change Martin indicates that he
is not speaking in his own voice.20 He seems to be commenting on the scene
described by Lilo. It does not matter who would speak like that in what
situation, and thus does not create a coherence problem. The following
utterance by Sabine is wonderfully related because Martin has focused on the
colour of the food and the contrast it makes on the wall. Starting in line 38
Sabine claims not to understand why people don't simply give children only
white things to eat. She solves the problem of the color contrast absurdly. This
is greeted by laughter. Martin thereafter confirms Sabine's lack of
understanding (41) by presenting yet another colour contrast (43).
In line 42 Conrad's fantasy moves further in the direction of black
humour: it can be inferred that plaster porridge would not contrast with a white
wall. Gerda augments the black humour (44); her suggestion would even make
feeding unnecessary. No one laughs at this, however. While it may be that
black humour is enjoyed in a rather "dry" manner (without laughter), perhaps
the humorous potential of the topic has simply been exhausted by this point.
Peter says, with a slightly complaining tone, that he has never been served red
beets since his childhood. The topic is gradually moved from a humorous
modality back to a serious one. Several persons have participated in the
amusing joint fictionalisations.
Both episodes have shown that it is completely unproblematic to insert
fantasies into discourse which are not yet worked out. The interlocutors carry
on the necessary coherence work by jumping from scene to scene. This
"loosening" of coherence is balanced by speakers' stylistic format tying
(speaking in headline style or using other elliptical constructions).
In Data 1 the amusing episode ends at line 35. There is a short pause, and then
Lilo explicitly introduces a new subject.
Coherence and Conversational Humour 141
In topic and keying Lilo returns to business talk. She asks for permission
to "interrupt the conversation". Peter agrees to that. The new topic is different
from the previous one, both in content and keying, which might make it
necessary to mark the shift explicitly. She wants to borrow "the Rotterengel", a
special dictionary she needs for interpreting. The group starts to talk about their
interpreting classes.
In Data 2 the topic is briefly maintained, but the keying is shifted into the
serious domain.
45 P: rote Rüben hab ich nie mehr gekriegt oder sonst was,
I have never gotten red beets or anything like that again,
46 auàer I woa jung.
except I was young.
47 I mein als Kind hob i roten Rübensalat gessen.
I mean as a child I ate red beet salad.
48 S: [(? ?)
49 ?: [(?also gut?)
okay well
50 E: ja, des hab ich aber, als Kind hab i des gemocht.
well, but I, as a child I liked that.
51 P: ja?
really?
52 E: jetzt kannst mich jagen damit.
now it scares me off.
53 P: i waaß a net. aba als Kind hob is gessen.
I don't know either. but as a child I ate it.
From line 45 on the group starts to seriously reflect on what they ate and
liked to eat as children, compared with the tastes they now have as adults. The
new topical aspect of taste differences between children and adults is
developed without creating comical perspectives and without integrating
laughter particles. Later the topic is dropped.
142 Helga Kotthoff
In the next data we witness how a topic is changed, but the keying remains
stable. The fictionalisation here serves to tease the interactant Gisela, who is
humorously razzed for not having bought enough soft pretzels for the party
held after their judo practice. Teasing requires that the conversational joking be
directed at someone present (Straehle 1993; Günthner 1996; Boxer and Cortés-
Conde 1997). Any deficit or shortcoming can serve as a reason for teasing.
Very often, the teased person is spoken about in the third person (tangential
address), the others present being the direct addressees of the amusing attack.
Teasings can be short or elaborated. Fictionalisations can represent a form of
elaborated teasing. The teased person is the protagonist of the construction.
Fritz (F), Gisela (G), Helmut (H), Klaus (K), Nadine (N), Oskar (O), Susanne (S),
Willi (W).
17 G: HAHAHAHAHA
18 H: ha is doch wahr,
well that's true.
19 soviel Butter das is doch wirklich mich gesund.
so much butter is really not healthy.
20 G: [Du hasch ja Butter im Gsicht.
[you've got butter on your face.
21 N: [HEHEHEHEHEHEHE
21 H: is doch wirklich nicht gesund so viel Butter.
it's really not healthy so much butter.
22 A: die viele Butter.
so much butter.
23 N: die Butter im Gesicht.
the butter on your face.
24 K HAHAHAHAHAHAHA
25 H; komm Butter. (wipes his face)
come butter.
26 (--)
27 H: ach nee, das sind ja die elektronschen Dinger, nich,
well no, those are the electronic things, aren't they,
28 combutter.
29 a: HEHEHEHEHEHEHE
30 ?:
Computer
computers
31 a: HEHEHEHE
32 A: Du bisch doch echt der letzte Vogel combutter.
you're really one of a kind. combutter.
6 Closing remarks
Humorous fictionalisations can fulfil many functions. A group can implicitly
negotiate shared norms and evaluations as in Data 1. People can also involve
themselves in playful expansions of absurd scenes as in Data 2 -just for fun. In
addition, fictionalisations can be used to tease other people. Here as well, we
find people participating in the creation of a comical scene. But in humorous
keying we also find counter-attacks; the teaser always runs the risk of
becoming the teased. Counter-teasings re-establish the humorous keying.
It is obvious that topics and keyings have separate identities. Humorous
keying can be coherently produced beyond topic boundaries, and topical
coherence as well can be maintained beyond the boundaries of keying. Keying
Coherence and Conversational Humour 145
Notes
1. Perspectivation is used here similarly to Graumann (1989) and Kallmeyer and Keim
(1996: 286), who claim that all human beings' perceiving and acting is done from a
specific viewpoint. A process of perspective setting and taking is always somehow
involved in verbal interaction. A speaker must reveal her/his perspective, at least to a
certain extent; recipients must show how they interpret the manifested perspective.
2. Coherence and discourse topicality are not equated here; see Fritz (1982) and Schegloff
(1990) on the differences.
3. See Verschueren (1995) on metapragmatics.
4. See on humorous interaction modalities also the articles in Kotthoff (1996b).
5. I use Koestler's (1964) terminology of bisociation of contexts, frames or perspectives
because it is a more general concept for explaining humour than the concept of opposing
basic lexico-semantic scripts used by Raskin (1985). An extensive discussion of the
concepts can be found in Kotthoff (1998 in preparation).
6. The principle of "least effort" is also found in the otlier relevance-tlieories mentioned
above.
7. With Hymes (1974) and Bauman (1986) we see performance as a multilevel activity.
Performances exhibit temporal brackets and frames; they aim at producing an emotive
effect.
8. Different tenninologies are used in the field of punchline humour; see Attardo (1994) and
Kotthoff (1998) for a discussion.
9. Complexities in the management of point of view, polyvalence of utterances, and poetic
effects, which demand more coherence work, are often either implicitly (Wienold 1983) or
even explicitly (Lászlo 1988) associated with literature.
10. In Kotthoff (1996b) various authors discuss the patterns inherent in humorous genres such
as mocking, teasing, joint ironising and personal anecdotes. See Sornig (1987) on fooling
around (German: blödeln).
11. One aim of the study was to differentiate different genres of conversational joking.
12. Naturally, there are also joint fictionalisations which are not humorous. Sometimes
speakers jointly embroider a journey they are planning or some wish they have.
13. Phettberg is a pseudonym which refers self-ironically to the bearer's corpulence. In
German Phettberg means literally "mountain of fat." The spelling "Phett" instead of "Fett"
is old-fashioned and is intended to imply a high-level of education.
14. "Phettbergs nette Leit Show."
15. All interlocutors speak Austrian German. The transcription takes account of this. Laughter
particles are noted wherever they occur, and the vowels in which they appear (HIHI or
HUHU etc.) are indicated.
16. See Couper-Kuhlen (this volume) for framing properties of prosody.
17. Children thereby learn to deal with colours and colour nuances. The numbers on the not-
yet coloured pictures are assigned colours which children are to use in filling in different
sections of the picture.
18. "Leit" corresponds phonetically to English "light" in the sense of a "light show," but as a
dialect term it corresponds to the standard term "Leute"/"people."
19. Without background knowledge we could not figure out the teasing dimension of Sabine's
remark. The teased do not necessarily defend themselves against the teasing.
20. In his essay "Footing" (1981), Goffman elaborated the possible division of roles within
speakers and listeners. He distinguishes degrees of responsibility for speaking and
Coherence and Conversational Humour 147
listening. The listener can be the addressee, the public or an eavesdropper. For Goffman,
"speaking" contrasts above all with "animating." This is, for example, the case with
quoted speech, where one puts one's own words into the mouths of other people and can
still manipulate the prosody so that in direct quotation a social type is simultaneously
created, as Couper-Kuhlen shows in her article in this volume. The animator presents
herself only as a sounding box for the words of others; Goffman further distinguishes the
principal, who bears responsibility, and the author, who is the creator of the text. See also
Günthner (1997) on reported speech.
21. Sacks (1972, lecture 5) showed that the best way to move from one topic to the next in
conversation is not by one topic closely followed by the next topic, but rather by what he
calls a stepwise move, which involves connecting what we've just been talking about to
what we're now talking about. He even writes "that what's thought to be a 'lousy
conversation' is marked by the occurrence of a large number of specific new topic starts
as compared to a conversation in which, so far as anybody knows, we've never had to
start a new topic, though we're far from whatever we began with, we haven't talked about
just a single topic; it just grew".
References
Attardo, S. 1994. Linguistic Theories of Humor. Berlin: de Gruyter.
Auer, P. and di Luzio, A. (eds). 1992. The Contextualization of Language.
Amsterdam: Benjamins.
Bateson, G. 1954/72. "A theory of play and phantasy". In Steps to an Ecology of
Mind. San Francisco: Chandler, 177-193.
Bauman, R. 1986. Story, Performance and Event. Cambridge: Cambridge University
Press.
Boxer, D. and Cortés-Conde, F. 1997. "From bonding to biting: conversational joking
and identity display". Journal of Pragmatics 27: 275-294.
Bublitz, W. 1989. 'Topical coherence in spoken discourse". Studia Anglica
Posnaniensia 22: 31-51.
Capone, A. 1995. "Dialogue analysis and inferential pragmatics". In F. Hundsnurscher
and E. Weigand (eds), Future Perspectives of Dialogue Analysis. Tübingen:
Niemeyer, 147-163.
Cook-Gumperz, J. and Gumperz, J. 1976. Context in Children's Speech. Papers on
Language and Context. {Working Papers No. 46). Berkeley: Language
Behaviour Research Laboratory.
Cook-Gumperz, J. and Gumperz, J. 1994. "The politics of conversation:
conversational inference in discussion". In A.D. Grimshaw (ed), What's Going
on Here? Complementary Studies of Professional Talk. Norwood: Ablex, 373-
395.
Couper-Kuhlen, E. This volume. "Coherent voicing: On prosody in conversational
reported speech".
Dascai, M. 1977. "Conversational relevance". Journal of Pragmatics 1: 309-328.
Dolitsky, M. 1983. "Humor and the unsaid". Journal of Pragmatics 7: 39-48.
148 Helga Kotthoff
Kallmeyer, W. 1979. "(Expressif) Eh ben dis donc, hein' pas bien': zur Beschreibung
von Exaltation als Interaktionsmodalität". In R. Kloepfer (ed), Bildung und
Ausbildung in der Romania, Vol. 1. München: Fink, 549-568.
Kallmeyer, W. 1981. "Gestaltungsorientiertheit in Alltagserzählungen". In R.
Kloepfer and G. Janetzke-Dillner (ed), Erzählung und Erzählforschung im 20.
Jh. Stuttgart: Kohlhammer, 409-429.
Kallmeyer, W. and Keim, I. 1996. "Divergent perspectives and social style in conflict
talk". Folia Linguistica 30: 271-299.
Koestier, A. 1964. The Act of Creation. London: Hutchinson.
Kotthoff, H. 1995. "Oral performance in interactional sociolinguistics: humorous
stories among friends". Toegepaste Taalwetenschap in Artikelen 52: 23-55.
Kotthoff, H. 1996a. "Impoliteness and conversational joking: on relational politics".
Folia Linguistica 30: 299-327.
Kotthoff, H. 1998. Spaß Verstehen. Zur Pragmatik von konversationellem Humor.
Tübingen: Niemeyer.
Kotthoff, H. (ed). 1996b. Scherzkommunikation. Beiträge aus der empirischen
Gesprächsforschung. Opladen: Westdeutscher Verlag.
Laszló, J. 1988. "Readers' historical-social knowledge and their interpretation and
evaluation of a short story". Poetics 17: 461-481.
Mulkay, M. 1988. On Humour. Its Nature and Its Place in Modern Society.
Cambridge: Polity Press.
Müller, K. 1992. "Theatrical moments: On contextualizing funny and dramatic moods
in the course of telling a story in conversation". In P. Auer and A. di Luzio
(eds), The Contextualization ofLanguage. Amsterdam: Benjamins, 199-223.
Norrick, N.R. 1989. "Intertextuality in humour". Humor 2: 117-139.
Norrick, N.R. 1993. Conversational Joking. Humor in Everyday Talk. Bloomington:
Indiana University Press.
Raskin, V. 1985. Semantic Mechanisms ofHumour. Dordrecht: Reidel.
Ruch, W., Attardo, S. and Raskin, V. 1993. "Toward an empirical verification of the
general theory of verbal humor". Humor 6: 123-136.
Sacks, H 1974. "An analysis of the course of a joke's telling in conversation". In R.
Bauman and J. Sherzer (eds), Explorations in the Ethnography of Speaking.
Cambridge: Cambridge University Press, 337-353.
Sacks, H. 1992. Lectures on Conversation. (Ed. by Gail Jefferson.) London:
BlackwelL
Schegloff, E. 1968. "Sequencing in conversational openings". American
Anthropologist 70: 1075-1095.
Schegloff, E. 1972. "Notes on a conversational practice: formulating place". In D.
Sudnow (ed), Studies in Social Interaction. New York: The Free Press, 75-119.
Schegloff, E. 1990. "On the organization of sequences as a source of 'coherence' in
talk-in-interaction". In B. Dorval (ed), Conversational Organization and its
Development. Norwood: Ablex, 51-78.
Schenkein, J.N. 1972. "Towards an analysis of natural conversation and the sense of
heheh". Semiotica 6: 344-377.
150 Helga Kotthoff
1 Introduction
In more than two decades of assiduous research into coherence, very little
attention has been paid to cases of coherence disorder or breakdown, i.e. of
what we propose to call disturbed coherence. Such neglect cannot readily be
explained, especially since disturbed coherence is by no means an infrequent and
unknown phenomenon in everyday communication. Quite the contrary: as
hearers we are used to appeal to the speaker with fixed phrases such as I'm lost
or fill me in or, more informally, hunh, whenever we are unable to reach a
coherent understanding of what we hear. And as readers we are arguably even
more often at a loss for how to make an essay, a poem, a newspaper
commentary, a technical manual coherent; and, to make matters worse in this
case, there are no writers vis-à-vis to whom we. could appeal to help us
negotiate the coherence of their texts.
In this paper, we would like to discuss disturbed coherence as a common
phenomenon of communication, which, on the one hand, must be kept distinct
from incoherent speech of persons whose command of language is impaired
(among them schizophrenics, cf. below), and, on the other hand, should not be
treated as simply the opposite of coherence. On the grounds that coherence is a
scalar notion, we argue that texts are frequently only partly coherent for their
hearers, or, to put it more strongly, that the hearer's understanding of coherence
can resemble but never totally match the speaker's. Thus, partial coherence is
not necessarily disturbed coherence. It turns into disturbed coherence at a point
when the extent to which the text is only partly understood is no longer
tolerated by the hearer. We distinguish between different types of disturbed
coherence (deliberate and accidental) and will present in this paper a detailed
account of some of its sources, such as topic drifts, topic changes, lack of
reference, frame and register breaks.
Our original intention was to keep within the confines of authentic and
attested spoken face-to-face discourse. However, relevant data is not easy to
154 Wolfram Bublitz and Uta Lenk
come by. To broaden our data-base, we therefore went hunting for examples in
all kinds of territories of discourse, ranging from available corpora like The
London-Lund-Corpus of Spoken English (LLC) to yet unavailable ones like The
Santa Barbara Corpus of Spoken American English (CSAE), as well as from
our own personal recordings of everyday talk to written fiction.
(1)
Observe this unit, gentlemen, which
(for want of a better word)
I call a self
It consists of an envelope enclosed by a void
and enclosing a void
Disturbed Coherence 155
(2)
ALMA: ...() look at this,
... isn't [this pretty]?
SUSAN: [these are] good.
... [2<X lime or the green X>2].
DICK: [2what is that2].
ALMA: .. [3Kerry always wears them3].
ROGER: [3bok bok3] bok bok bok.
ALMA: ... She brought some,
ROGER: bok [bok].
ALMA: [I al]ways comment [2on them2].
SUSAN: [2well you2] have to be
nice to me,
[3if you want to talk3][4me in4][5to shar5][6ing
my candyó].
JESS: [5<VOX %uh
VOX>5],
ALMA: [3She brought some3] «
DICK: [3aw=3],
[4that's4][5cute5] [6Kerry6].
JESS: ... [7<VOX %uh VOX>7],
ALMA: [7 X <X the table for them X>7].
BOB: [7it's not7] cute at all,
X: [8 XX <X get yourself X>8].
BOB: [8it's very elegant8].
JESS: I am,
.. I am not mean to you all the time.
Disturbed Coherence 157
4 Some examples
It is precisely because he continues to operate on the default assumption of
coherence that the hearer appeals to the speaker for elaboration and
clarification, i.e. to help him overcome the disturbance and find coherence. To
this end, he resorts to fixed phrases such as you 've lost me, what do you mean, I
do not understand, fill me in, how does that tie in with... Here is one example
(for more, cf. below):
(3)
S: Who never stopped talking or something - urn - but we
really need to talk about this.
D: About what?
158 Wolfram Bublitz and Uta Lenk
(4)
MARILYN: I-1 really,
... we read this great book.
.. called,
.. The End of Nature?
PETE: .. Oo=
@ [Sounds wonderful].
Disturbed Coherence 159
(5)
b [...] I've been thinking [...] over writing this
dictionary [...] and looking [...] at the work of
slightly older people [...] - all their examples and ■
half of them are military -[...] they do escort in
terms of warships I do it in terms of red roses - -
A o\h - m\
b again and again and again [...]
A m\
b (do you) see what I mean
A \m (coughs .)
b it's the collocaters that have changed
A ye\s
(LLC 1.10.226-235)
(6)
a [...] the meaning ofthat little diagram is that
everybody's got to do the central three he's got to
pick another seven on top of that and he's got to pick
up his seven by following the paths -
A *how do you m\ean he (he)* +m+
a *and if he wants* if he wants to take say this one
+he's+ got to do everything in bet**ween**
A **oh** I\ see . ye\s .
a so that if he wants to do this this commits him already
to seven three four five *six seven*
A *I f\ollow* y\es . y\es
a so that he can't really do that and that - -
A no I f\ollow - . I f/ollow [...]
(LLC 2.2a.376-388)
(7)
Swiss pride must be deserved. Salu K..., I am the nun. If that's enough, you are still
his. That is a brave cavalier, take him as your husband. Karoline, you well know,
though you are my Lord, you were just a dream. If you are the dove-cote, Mrs. K. is
still beset by fear. Otherwise I am not so exact in eating. Handle the gravy carefully.
Where is the paint-brush? Where are you, Herman?
(Roberts and Kreuz 1993: 456f.)
Listening to this (or, indeed, reading it), you would most certainly regard
(7) as a piece of totally incoherent discourse. However, if we had listed (7) as
an example of disturbed coherence - not telling you that it was produced by a
schizophrenic - or appealed to your default assumption of coherence, your
reaction would most likely have been very different. You would have suspended
your judgement until it had dawned on you that it was not for a lack of data that
you failed to infer the missing links (and thus coherence), but that there were no
links missing because the speaker either had not intended to or else had been
unable to produce coherent discourse in the first place. The hearer who has
come to such a conclusion will no longer operate on the default principle of
coherence. He will not give the speaker, or rather the text, the benefit of making
sense and will refuse to further engage in cooperative interaction. So the
question to ask, then, is not whether or not a text like (7) is coherent but
whether or not its hearer or reader is willing to assume that it is coherent, i.e.
whether or not he is willing to make it coherent. This is entirely his choice (as
well as in accordance with the hearer-knows-best principle of describing
coherence), and is independent of the cause of the speaker's failure to secure
coherence; she may, for example, be schizophrenic or aphasic, too young or too
old, drunk or drugged.
The explanation above accounts for those cases of incoherence that are
caused by some kind of impairment. But it also helps us to understand why a
reader, after having failed to understand its coherence, may end up concluding
that some literary text is incoherent (indeed, as incoherent as schizophrenic
talk), because its writer did not abide by the default principle of coherence.
There are several reasons for the reader's readiness to come to some such
conclusion (unfamiliarity with the genre or the author, insufficient interpretative
competence among others). Other readers, for instance those who have greater
interpretive skills or who have had a chance of conversing with the writer and
thus of negotiating meaning, force and coherence (i.e. of establishing a common
context) might find the same text to be either fully coherent or might regard it
merely as a case of (more or less) disturbed coherence.
162 Wolfram Bublitz and Uta Lenk
(8)
A ich glaube ich bin gegen Sesam allergisch, ich hab
nämlich neulich - oder gegen Mohn. ich hab -
B grüne oder weiße?
A hä?
B grüne oder weiße?
A grüne oder weiße?
B grüne oder weiße Bohnen?
A Mo = =hn!
(Uta Lenk, personal data, March 1997)
'A I think I am allergic to sesame seed. the other day I - or to poppy seed. I had -
B green or white?
A hunh?
B green or white?
A green or white?
B green or white beans?
A po==ppy seeds!'
(9)
When it is Frank's turn to testify, the bailiff try to swear him in. 'Do you swear to
tell the truth, the whole truth, and nothing but the truth?' 'I'll take the first one,' say
Frank. [...] 7 do will suffice,' say the Bailiff. 'Why would you give me a choice if I
Disturbed Coherence 163
got to choose all three?' ask Frank. The appropriate reply is I doj the bailiff say. 'If
I say I do, I'm liable to wind up married to somebody. How about Okay?
'Whatever.' 'Let me hear those choices again?'
(W. P. Kinsella. 1994. "Bull." In: Brother Fran's Gospel Hour. Toronto:
HarperCollins, p. 9.)
(10)
A While I was taking Chinese, during my year as teaching
assistant, I was planning to go to China later on, to
really pick up the language there. But that was the
year of the Tien An Men massacre, and that kind of
changed my plans. I -
164 Wolfram Bublitz and Uta Lenk
Speaker A's reference to the Tien An Men massacre opens the frame of a
totalitarian regime where she would not want to spend any amount of time to
learn the language. C's statement that the massacre never happened is
completely incoherent for A because it negates the existence of one particular
component of this historical frame, an existence which she presupposes to be
true as it is well-known and widely documented. This is totally unexpected and
so unbelievable that it triggers a question for clarification instead of a clear
contradiction.
And there is a third group of motives that are neither friendly nor
unfriendly. The speaker may, for instance, deliberately disturb coherence simply
in order to get the hearer's attention.7 Or she may be deliberately vague,
ambiguous, non-specific, thus preventing the hearer from an easy, obvious and
immediate ascription of coherence, because she wants to wait for the hearer's
reaction, for his interpretation of coherence. She leaves the question of
coherence open, as it were, leaves it to the hearer to answer it. Such behaviour
supports our view that participants accept mutual responsibility for making a
text coherent (in her workshop contribution, Kotthoff gave examples where
deliberate misunderstandings result in temporary disturbances of coherence and
which are employed as means of conversational humour).
To conclude, the speaker's motives when deliberately instrumentalising
disturbed coherence are opportunistic but she never aims at a total and
permanent breakdown of coherence. Incidentally, for the hearer who fails to
ascribe coherence it makes no difference whether or not the disturbance
happens accidentally or is deliberately caused by the speaker.
The topic of our paper is not 'disturbed cohesion' or, indeed, 'incohesiori.
Taking a text and stripping it of its cohesive means could be described as
making it 'incohesive'. As there is no direct and mandatory correlation between
Disturbed Coherence 165
For the hearer, the act of identifying a referent can fail or can be partly or fully
successful. It can be immediate or delayed and can take a shorter or longer span
of time depending on how much information is readily available and how much
cognitive effort is ultimately required. These options correspond to different
degrees of disturbed coherence. In (11), the coherence of B's first utterance is
strongly disturbed for d, who completely fails to identify the referent of feet:
(11)
A the thing to do is obviously to swing the PhD subject
round to something -. nearer what I'm being paid to do
*you see*
B *so you've* gone off feet -.
d you've what
166 Wolfram Bublitz and Uta Lenk
A,B (- - - laugh)
B he's gone off *feet*
A Tm* no longer Peter's footman (- - +-+ laughs)
B +(-laughs)+ you've heard of the fo/ot [...]
d no
B Abercrombie's feet -. *rhythm . rhythm*
d *well yes but « n o t 1 syll»* fill me in yeah
B well he'd been doing a thesis on feet
d oh gosh yes of course
(LLC 2.4.1065-1083)
Topic drift occurs when the global topic is temporarily neglected or even
permanently abandoned. Discourse then develops according to what is
associatively closest or "easiest to say next" (Hobbs and Agar 1986: 231) rather
than to what the speaker's projected goals of the global topic demand. This
complies with the hearer's view that for the interpretation of the current
utterance "the immediately preceding utterance has a special significance
because it is against the mutual cognitive environment as established by this
previous utterance that the oncoming utterance is going to be interpreted"
(Jucker 1993: 72). Each single step of slightly shifting the topic will not affect
Disturbed Coherence 167
(12)
a you were telling us a . a long complicated story about
Eileen's sons last night -I hadn't quite got them in
order
B well she has four boAys
[...7 lines leading up to their names]
B BeVn Don Luke and Co\in
[... 88 lines, in which B starts to talk about Eileen having moved into a rather
outback kind of place and then, in long monologues, drifts from topic to topic;
when finally she comments on the immense costs you have to invest in dental
chairs, a steers her back to the original topic]
a and how does this tie in with Ben
(LLC 1.13. 1ff)
(13)
Student: wär's möglich daß sie [seine Hausarbeit] eher
synchron nicht diachron ist
Professor: ja kein Problem
Student: mit synchron hob ich's nicht so
Professor: ja ja *aber Sie* könnten vielleicht
Student: *äh diachron*
(Wolfram Bublitz, personal data, April 1997)
(14)
A well I haven't seen her this te\rm
B I \have I've only seen her o\nce since I he\ard this
news I think in in the in the refe\ctory and I didn't
actually spe\ak to her — I had a seminar today in
which . people hadn't read the stu\Aff because [...]
(LLC 1.4.1076ff)
(15)
c he's fast asleep -
B i/she
c yes -.
B oh go\od . Jean Piage\t what a what's the point of
having a book about hi\m around
(LLC 2.10.326fi)
For the participants actually present, however, these abrupt topic changes,
though definitely locally incoherent, do not lead to disturbed coherence for
various reasons. The new topic introduced after the topic change in (14) is
coherent on a global level because, like the preceding topic, it contributes to the
global topic 'exchanging university news'. Therefore, its introduction is not a
violation of the participants' expectations triggered by the global topic. The
local incoherence in (15) is likewise of no concern for the global coherence,
though for a different reason. There is not one superordinate global topic as in
(14) but both the preceding and the following topic in (15) relate to the general
conversational purpose or goal of maintaining and developing social contact.
Here, the exchange of information on the cognitive or ideational level is
secondary to the exchange of information on the interpersonal level. Analysts,
when looking only at truncated sections of some such conversations, may not be
Disturbed Coherence 169
aware of this, but those participating in the interaction check all incoming
messages, including those after an abrupt change of topic, against the main
conversational goal of making them coherent.
Abrupt topic changes can, however, prompt disturbed coherence.
Example (16) illustrates a case where both the next topic and the topic change
itself lead to disturbed coherence.
(16)
PETE: Yeah.
.. That would be good.
... Cause [all] of that stuff should go into the
compost pile to begin with.
MARILYN: [X]
..Yeah.
..(H) Actually,
you know,
.. Zeke the sheik.. is a local.
...You know,
the guy whose compost pile blew up?
PETE: ... Oh no I don't know a[bout this].
MARILYN: [Didn't you hear] about him?
PETE: No.
(CSAE: Conceptual Pesticides)
Marilyn takes up Pete's cue of compost pile to initiate a topic drift from
gray water systems, a subtopic of their overall topic environmental issues, to a
specific case of compost-pile-trouble as another subtopic. The abrupt change of
topic is introduced by dropping a name and commenting on its owner {Zeke the
sheik is a local). Pete's lack of reaction ('...' indicates a pause) is recognised as a
sign of his failure to make this utterance and thus the topic change coherent.
Obviously, he is unable to detect a superordinate global topic (though, in fact,
there is one) to which he could relate the utterance at the moment of hearing it.
Nor is he able to relate the new information to a general conversational goal. At
this point in the conversation, the utterance therefore comes totally unexpected
and disturbs the coherence of the ongoing discourse. This is recognised by
Marilyn, who then gives additional details in order to help Pete identify the
referent, thus enabling him to see the connection between the new and the old
topic and ultimately find coherence.
To conclude, abrupt topic changes are always locally incoherent but
coherent on a global level provided the new topic is a noticeable contribution
either to a superordinate topic or to a general conversational goal on the
170 Wolfram Bublitz and Uta Lenk
(17)
c the only diplomatic way out is to change the subject
very interestingly -
d we're putting up a lot of shelves in the other room
(LLC 2.10.1019f)
(18)
A: im German Department in Pittsburgh hat jemand eine Dissertation über
Aschenputtel geschrieben *und*
B: */Aschenputtel*
C: */worüber*
(Wolfram Bublitz, personal data, January 1997)
When we gave the oral presentation on which this paper is based, we started
our talk with You know, the funniest thing happened to me this morning on my
way to our workshop. In more than two decades of assiduous research into
coherence ... We deliberately chose this onset to create a feeling of disturbed
coherence in our audience. Apparently, we were partly successful, and this was
possible because You know, the funniest thing happened ... is a typical story
"preface" (Sacks 1974: 340 ff) but not an appropriate preface of an academic
talk. It is in the wrong register and opens the wrong genre. Generalising the
default principle of coherence, we can say that as long as there is no reason to
assume otherwise participants in a conversation assume that the current
situation, frame, register and genre still hold (cf. Widdowson 1983: 45). A
noticeable deviation from their conventions disturbs the hearer's search for
coherence. Unfortunately, we cannot go into this aspect in detail here but will
leave it for further research.
172 Wolfram Bublitz and Uta Lenk
8 Conclusion
Contrary to the general impression created by much research into coherence
disorder, it is not only schizophrenic talk that is incoherent. In our research, we
found disturbed coherence of a different kind in everyday face-to-face
conversations. This did not come as a surprise because, after all, ascribing
coherence is a highly idiosyncratic and thus easily disturbed activity. Cases of
disturbed coherence are, nevertheless, not abundant in conversations for two
reasons: speakers and hearers alike operate on a general default principle of
coherence and they accept mutual responsibility for collaboratively achieving
coherence. For the hearer, the coherence of a text is disturbed when he is unable
to make it coherent but assumes that it could be made coherent because he has
no reason to believe otherwise. Starting from the assumption that coherence is a
scalar notion, we argue that coherence is disturbed when the extent to which a
text is only partly coherent is no longer tolerated by the hearer. The hearer's
failure to make a text coherent is, inter alia, prompted by violations of
expectations as they appear in cases of topic drift, topic change, unclear
reference, frame- and register-breaks.
Notes
1. In this paper, we use she to refer to the speaker and he to the hearer.
2. For a similar view cf. e.g. Dahlgren as quoted in Roberts & Kreuz (1993: 455).
3. The following transcription conventions appear in the CSAE examples: a carriage return
indicates the end of an intonation unit, a comma stands for a continuing intonation
contour, a period indicates a final intonation contour, a ? stands for appealing intonation
contour; two dashes (—) indicate a truncated or incomplete intonation contour after (self-)
interruption; [ ] indicate speech overlap (numbered consecutively when several overlaps
occur close to each other); .. mark a short pause,... a medium pause and ...() a pause that
will be timed in the published version of the corpus; = marks the lengthening of the prior
sound; (H) is inhalation, (Hx) exhalation, @ stands for a syllable of laughter; <X word
X> indicates uncertain hearing, X an indecipherable syllable.
4. Ciliberti (this volume) shows that in certain situations, such as e.g. examinations, the
default principle is temporarily suspended. Typically, these are kinds of interactions
where all participants know that the relevant rules for this particular situation require the
suspension.
5. Unless the digression constitutes out-of-topic talk or is a so-called situational digression,
cf. Bublitz (1988: 94ff) and Lenk (1997).
6. For "in fact, one criterion for a diagnosis of schizophrenia is incoherent speech
(American Psychiatric Association 1987)" (Roberts and Kreuz 1993: 456). However,
even schizophrenic talk is not as uniformly incoherent as is often argued but may be
more coherent or less coherent. This is pointed out by Roberts and Kreuz, who have also
Disturbed Coherence 173
observed that "schizophrenics may be considered more coherent when they are assumed
to be speaking coherently" (456).
7. Occasionally, speakers even tell the hearer in so many words that what they are saying
does not fit. Or the incoherence may be the (intended) result of a clash of maxims: the
speaker may deliberately observe the politeness maxim at the expense of the maxims of
relevance or quality.
8. It is very interesting that, if several speakers are involved in such a process of teasing,
often a kind of agreement between the speakers as to the aim and method can be noticed
despite the fact that their intention was never verbalised.
9. "Local coherence" here means that two successive utterances are linked by coherence
relations such as "temporal succession, causal relation, explanation, semantic
parallelism, elaboration, exemplification, contrast, background [...]" (Hobbs and Agar
1986: 220f); cf. also Lenk 1998 and Sanders and Spooren (this volume).
References
Bublitz, W. 1988. Supportive Fellow-Speakers and Cooperative Conversations.
Amsterdam: Benjamins.
Bublitz, W. 1994. "In the eye of the beholder: 'the rather mystical notion of
coherence'". In K. Carlon, K. Davidse and B. Rudzka-Ostyn (eds), Perspectives
on English. Studies in Honour of Professor Emma Vorlat. Leuven, Paris:
Peeters, 213-230.
Bublitz, W. 1996a. "T bought a freezer - you've bought a freezer - they've bought a
freezer': repetition as a text-building device". In C. Bazzanella (ed), Repetition in
Dialogue. Tübingen: Niemeyer, 16-28.
Bublitz, W. 1996b. "Semantic prosody and cohesive company: somewhat predictable".
Leuvense Bijdragen 85: 1-32.
Chafe, W. 1994. Discourse, Consciousness, and Time. The Flow and Displacement of
Conscious Experience in Speaking and Writing. Chicago: Chicago University
Press.
Clark, H.H. and Wilkes-Gibbs, D. 1992. "Referring as a collaborative process". In P.
Cohen, J. Morgan and M. Pollack (eds), Intentions in Communication.
Cambridge, MA: MIT Press, 463-493.
Halliday, M.A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Hobbs, J. and Agar, M. 1986. 'The coherence of incoherent discourse". Journal of
Language and Social Psychology 4: 213-232.
Jucker, A.H. 1993. 'The structure and coherence of discourse". In H. Löffler, C.
Gromilund and M. Gyger (eds), Dialoganalyse IV Referate der 4.
Arbeitstagung Basel 1992. Tübingen: Niemeyer, 71-78.
Kotthoff, H. This volume. "Coherent keying in conversational humour. Contextualising
joint fictionalisation".
Lenk, U. 1998. Marking Discourse Coherence: Functions of Discourse Markers in
Spoken English. Tübingen: Narr.
174 Wolfram Bublitz and Uta Lenk
Roberts, R.M. and Kreuz, R.J. 1993. "Nonstandard discourse and its coherence".
Discourse Processes 16: 451-464.
Rochester, S. and Martin, J.R. 1979. Crazy Talk: a Study of the Discourse of
Schizophrenic Speakers. New York: Plenum Press.
Sacks, H. 1974. "An analysis of the course of a joke's telling in conversation". In R.
Baumann and J. Sherzer (eds), Explorations in the Ethnography of Speaking.
London: Cambridge University Press, 337-353.
Sanders, T. and Spooren, W. This volume. "Communicative intentions and coherence
relations".
Schegloff, E. 1990. "On the organization of sequences as a source of 'coherence' in
talk-in-interaction". In B. Dorval (ed), Conversational Organization and its
Development. Norwood, NJ: Ablex, 51-78.
Seidlhofer, B. and Widdowson, H. This volume. "Coherence in summary: the contexts
of appropriate discourse".
Widdowson, H.G. 4 1983. Teaching Language as Communication. Oxford: Oxford
University Press.
Coherence and Misunderstanding in Everyday
Conversations
(1)
1 A. cos 'avete fatto ieri sera f
2 B. abbiamo visto il toro alla tele ↓
3 A. e cos 'ha fatto il toro T
4 B. (-)ma no(-)lLTOROi
Normally, for each turn, more than one interpretation would be suitable and
coherent, but the one intended by the speaker must be recognised. In other
words, the interlocutor has to reconstruct the interpretation and the coherence
as they were intended by the speaker, with a degree of tolerance that differs
from one conversation to another, partly depending on the goals of the speaker
and of the exchange itself. If coherence can be a requisite for correct
understanding, in some cases the interlocutor can fail in this task, assigning to a
turn a coherence that differs from the speaker's intended coherence, and an
interpretation that differs from the speaker's intended meaning. In these cases
we are faced with a misunderstanding, which we propose to characterise on the
basis of two main features (cf. Weigand forth.):
• the mismatch between the speaker's intended meaning and the interlocutor's
interpretation of a given turn,
Coherence and Misunderstanding 177
2.1.1 Mismatch
The participant who fails to come up with the expected interpretation of a turn
normally succeeds in reconstructing another kind of coherence between her/his
'wrong' interpretation and the previous conversation. Up to then, we can
assume that the participants' views of the ongoing conversation have matched,
at least as far as the main points are concerned. As a consequence of the
misunderstanding, the speaker's coherence and the interlocutor's coherence no
longer match, thus creating two different contexts of interpretation for the turns
to come. After the misunderstanding has occurred, the interlocutor's reaction to
the speaker's next turn will be coherent with her/his own interpretation of the
dialogue, which no longer matches the other's. In other words, the interlocutor
who has misunderstood will behave coherently with her/his misunderstanding,
but not in a manner appropriate to the speaker's intentions. Adopting a
geometrical metaphor, we could say that, if before the occurrence of a
misunderstanding the participants' views of coherence in conversation
coincided, afterwards they become like parallel lines, eventually diverging
perceptibly, as is the case in the following fragment which took place in a
context of doctor/patient interaction. In this episode, while the patient was
dressing behind a curtain, after an x-ray examination of her hip-bone, the doctor
was examining the x-ray outside, and noticed the presence of a metal paper clip
in the patient's abdomen. All at once, the doctor asks, without showing her the
x-ray:
(2)
1 M. scusi (-) ma lei mangia graffe T
2 P. ma (-) si (-) ogni tanto (-) se capita↓
3 M. ma come (-) cosa vuol dire ogni tanto (-) quando le capita f
guardi che le fanno male ↓
(-)
4 P. si (-) lo so che fanno male (-) ma infatti ne mangio una
volta ogni tanto ↓
5 M. si (-) ma sarebbe meglio se non ne mangiasse del tutto ↓
6 P. ma non è che tutti i giorni io mangio krapfen ↓
7 M. ma cosa dice t (-) io ho detto graffe↓ne ha una nell 'addome ↓
realignment
clash
parallel
discourse
alignment
I interlocutor M interlocutor P
Figure 1. Diverging coherence in example 2
180 Carla Bazzanella and Rossana Damiano
(3)
1 B. con un assegno no <??>
2 C. io ho il conto qua
3 B. ha il conto qua quattro otto sei? duecento lire di commissione
4 C. ha bisogno del numero del conto # quello gia' e'piu' difficile
[silenzio]
5 C. ecco # /'/ conto e' nove tre due otto barra uno
6 B. come e'intestato?
7 C. non e'intestato
8 B. no come e' intestato il conto
9 C. ah XYZ Francesca
2.1.3 Causes
The role of ambiguities in generating misunderstanding has been underlined by
several scholars (cf., among others, Blum-Kulka and Weizman 1988; Zaefferer
1977), and seems to be confirmed by the examination of the data: although
there are instances of misunderstanding in which no ambiguity is present,
ambiguity causing a misunderstanding was found in approximately 70 % of the
instances considered. In example (4), again taken from our corpus, the
referential ambiguity of the subject of the verb phrase "they don't exist", in the
first turn, reflects on the understanding of the sentence topic. After the
misunderstanding occurs in line 3, the speaker makes it clear, in line 4, that he
was talking about limits and not products. Interestingly, it is difficult, here, for
the interlocutor to accept such an interpretation, as is shown by her surprised
reaction. Nevertheless, the 'wrong' interpretation displays cohesion with the
first sentence, since 'they' could also refer to 'the products' instead of 'the
limits', even though the latter is preferred as the nearer.
(4)
1 A. ci sono una serie di prodotti per i quali noi non diamo limiti perché
ufficialmente non ci
2 sono
3 B. come non ci sono↓(-) non ci sono i prodotti (-) ufflcialmente T
4 A. non i prodotti (-) i limiti↓
182 Carla Bazzanella and Rossana Damiano
'1 A. there's a series of products for which we don't give limits as they don't
exist
2 officially
3 B. what do you mean don't exista↓(-) the products don't exist
(-) officially↑
4 A. not the products (-) the limits↓'
with the interpretation context constructed until that point. Bublitz (1989)
argues that for the ascription of coherence to an utterance to be confirmed it has
to match with subsequent data from the same speaker. With regard to
computational linguistics, McRoy and Hirst (1995) have designed a default
system that exploits preferences based on adjacency pairs to evaluate the
coherence of new turns and to detect misunderstandings.
Real data evidences that the task of reconstructing what the 'wrong'
interpretation consists of is normally a simple one for human participants,
probably because they are aware of the weak spots (ambiguities, speaker's
failures, etc.) of their past contribution to the conversation and of their own
interpretation of their partner's turn. As speakers, in particular, they seem to
have an immediate understanding of what has gone wrong and their repair turns
are thus explicitly aimed at correcting the faults detected in their partners'
interpretation. Moreover, the participant who tries to reconstruct how the
other's interpretation differs from her/his own (no matter whose fault it is) is
guided in this task by the coherence constraints of both the turns that come
before the misunderstanding and the turns that follow it. In this sense, the whole
context that surrounds the misunderstanding is not only meaningful for the
interlocutor's beliefs and intention, but also because it provides the repairing
participant with a valuable guide for the task of reconstructing the alignment
between the participants.
We will deal here with four aspects of the process of repair which occurs after a
misunderstanding:
• the agent of repair
• the phases of negotiation
• the devices of repair
• the placement of repair.
When coherence can no longer be assigned, the participants must work
together to reconstruct the lost common ground essential to successful
communication, eventually abandoning their 'individual' visions of coherence.
In this sense, coherence in a conversational and interactional context can never
be said to be static or ascertained, and, even when it is treated as if it was, it can
suddenly turn to uncertainty and need further work for its re-establishment9.
The participant who realises that a misunderstanding has occurred,
whatever her/his role in the conversation, first has to signal the difficulty, so that
she/he can act together with her/his partner to re-establish the lost common
coherence. This result is obtained by means of a negotiation, open to different
184 Carla Bazzanella and Rossana Damiano
endings. The form this negotiation normally assumes is the following (cf. Figure
2).
Here, we are dealing with the second step: cDoes the participant who has
detected the misunderstanding make a repair?'10. The participant who has
detected the misunderstanding can make a repair or not (second step). If she/he
does not, a communication breakdown is likely to happen.
If the participant who carries out the repair turn proposes a diagnosis of
what has happened and a consequent solution, her/his partner can confirm or
disconfirm the proposed reading (third step): cf. line 9 in 3, where the discourse
marker oh underlines the interlocutor's uptake.
If the interlocutor refuses the repair turn, there can be a conversational
breakdown, or a shift in conversational topic if they decide to disregard it and
go on (see third step, b). By contrast, if she/he accepts it completely (see third
step, a), we have a 'fresh start'; if the acceptance is partial, participants enter a
cycle of further negotiation.
The effectiveness of repair turns in human conversation can be explained
as a consequence of the participants' intuitions about lost coherence (as we said
earlier, cf. 2.2). There is, in fact, a strong relationship between a participant's
intuition about the misunderstanding and her/his repair. The speakers, in
particular, after reconstructing what has triggered the interlocutor's 'wrong'
interpretation and coherence, usually try to indicate the intended coherence
more clearly, not only by simply correcting, as they sometimes do, but also by
stressing or reinforcing the relevant features in their misunderstood turn. This
aim is achieved in several ways, depending on the factors that have triggered the
misunderstanding: by a more or less literal repetition, when the speaker thinks
that the interlocutor did not catch her/his utterance properly, by specifying what
the correct alternative was if there was ambiguity, or by explicitly contrasting
the interlocutor's interpretation, when it is easily reconstructed, with her/his
intended meaning (cf. 4).
With regards to the placement of the repair, the interlocutor's incorrect
interpretation is in most cases detected immediately afterwards by the speaker,
thanks to her/his inappropriate reaction, and corrected by the speaker in her/his
next turn. This case corresponds to 'third turn repair' in Schegloff's (1992)
terms.
More rarely the interlocutor her/himself, thanks to the speaker's next
contributions to the conversation, realises she/he has misunderstood and
corrects her/himself (fourth turn repair).
These are the most common patterns, but they can be extended over
several turns, so that repair can take place virtually in any of the turns that
follow the misunderstood one, starting from the third one, as shown by our
data, where instances of repairs can be found at the twelfth, fifteenth, and
thirtieth turn.
186 Carla Bazzanella and Rossana Damiano
3 Conclusions
We have briefly dealt with three aspects of the complex relation between
coherence and misunderstanding:
• the mismatching of speaker's and interlocutor's coherence as the source of
misunderstanding;
• the awareness of the lack of coherence as the start of a repair process;
• the repair of misunderstanding and the re-establishment of 'shared'
coherence.
The phases of this interactional cycle, based on negotiation and sequential
processes, have been analysed on the basis of real data, and some linguistic
devices (among the several involved in this process) have been stressed.
Within this framework it is evident that understanding itself, let alone
misunderstanding, is not a clear cut notion, but a product of coherence
ascription and, when necessary, of negotiation.
Notes
1. Our 'corpus' includes 63 instances of misunderstanding, partly extracted from existing
text corpora, partly collected for the specific purpose. Two of the existing corpora are
published and two smaller ones are unpublished. In the first group, Mauro, Mancini,
Vedovelli and Voghera 1993 is a collection of excerpts taken from both telephone and
face-to-face conversations, in different contexts that range from formality to informality,
while Gavioli and Mansfield 1990 consists of transcriptions of complete book-shop
encounters. The two unpublished corpora were collected by Cristina Ferrus and Orsola
Fornara for their theses in the Philosophy of Language (1994, University of Turin, Italy).
2. Usual transcription conventions are adopted in the corpus we collected; conventions of
other corpora are maintained.
3. See Bazzanella and Damiano 1997 for a proposal for classification of the levels of
misunderstanding.
4. The second feature, in particular, rules out instances of planned or intentional
misunderstanding and exploitation of misunderstanding (cf. Souza 1985).
5. It is worth noting that other interlocutors, who were listening to this conversation in the
adjacent room, realized immediately that a misunderstanding was occurring.
6. The reason why this kind of failure is not always immediately recognizcd is that thc
mechanism sketched above is not absolutely safe from failures itself, since every new turn
is open to a range of different interpretations, thus allowing for the misalignment
occasionally to remain covert for several turns.
7. In fact, if the interlocutor has to make an effort to come to the correct understanding, the
speaker also makes a complementary effort to show her/him the way. Preference
mechanisms, like adjacency pairs, topical coherence or different types of cohesion
(repetitions, collocations, discourse markers, etc.) have been indicated by researchers as
Coherence and Misunderstanding 187
examples of the devices used by the speaker to obtain this aim, but, in spite of these
devices, misunderstanding occurs and is dealt with interactionally.
8. These devices are often accompanied by discourse markers, which serve to communicate
the interlocutor's uncertainty and to make it a mutual goal to establish the compre
hension the interlocutor is not able to reach on her/his own.
9. In monological texts, too, a re-interpretation is often necessary, cf. Conte this volume.
10. We are not interested here in step one, where misunderstanding is not detected, and the
interaction goes on without any changes.
References
Bazzanella, C. (ed). 1996. Repetition in Dialogue. Tübingen: Niemeyer.
Bazzanella, C. and Damiano, R. 1997. 'Il fraintendimento nelle interazioni quotidiane:
proposte di classificazione". Lingua e Stile XXXII: 173-200.
Bazzanella, C. and Damiano, R. (forthcoming). "The interactional handling of
misunderstanding in everyday conversation". Journal of Pragmatics (special
issue on Misunderstanding).
Blum-Kulka, S. and Weizman, E. 1988. "The inevitability of misunderstandings:
discourse ambiguities". Text 8: 219-241.
Bublitz, W. 1989. "Topical coherence in spoken discourse." Studia Anglica
Posnaniensia XXII: 31-51.
Conte, M.-E. 1988. Condizioni di Coerenza. Ricerche di Linguistica Testuale.
Firenze: La Nuova Italia.
Gavioli, L. and Mansfield, G. 1990. The PIX1 Corpora. Bologna: Cooperativa Libraria
Universitaria Editrice Bologna.
Grice, P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University
Press.
Mauro, T. de, Mancini, F., Vedovelli, M. and Voghera, M. 1993. Lessico di Frequenza
dell'Italiano Parlato. Milano: Etas.
McRoy, S. and Hirst, G. 1995. "The repair of speech acts by abductive inference".
Computational Linguistics 21: 435-478.
Reilly, R.G. (ed). 1987. Communication Failure in Dialogue and Discourse.
Amsterdam: Elsevier.
Schegloff, E.A. 1992. "Repair after next turn: the last structurally provided defense of
intersubjectivity in conversation". American Journal of Sociology 97: 1295-
1345.
Souza, F. de. 1985. "Dialogue breakdown". In M. Dascal (ed), Dialogue: an
Interdisciplinary Approach. Amsterdam: Benjamins, 415-426.
Weigand, E. (forthcoming). "Misunderstanding: the standard case". Journal of
Pragmatics (special issue on Misunderstanding).
Zaefferer, D. 1977. "Understanding misunderstanding: a proposal for an
explanation of reading choices". Journal of Pragmatics 1: 329-346.
The Effect of Context in the Definition and Negotiation of
Coherence
Anna Ciliberti
Università per stranieri, Perugia
The oral examination in Italian universities is the last stage in a complex and
diversified didactic process during which the student experiences several
didactic modes and comes into contact with different types of texts. This
examination may assume different forms depending upon factors related to the
examiner, to the candidate, and to the subject taught.
As to factors related to the examiner (in an asymmetrical situation like the
oral examination), the examiner's individual traits of personality, age, gender,
geographical and social origin, together with the subject taught and her
pedagogic stance, play an important role in determining the way in which the
examination will be conducted. Among this set of variables, a special place is to
be ascribed to the examiner's "pedagogic voice" - as Bernstein (1990: 190) calls
it - i.e. to "the distinguishing features of specialised pedagogic communicative
practices".
190 Anna Ciliberti
(1)
Complainant: You know and all of a sudden he said 'old drunken bitch' I said
What' re you talking about' I said. We're together we gotta help
each other' you know
Police officer: —» What's your date of birth?
The interactional sequence is thus the following: (i) the student introduces
a topic (turn 1); (ii) the examiner interrupts the student and expands the topic
(turn 2); (iii) the student tries to go on with what she was saying disregarding
the last part of the examiner's turn (turn 3); (iv) the examiner takes no notice of
the student's topic resumption and brings talk to an end (turn 4).
192 Anna Ciliberti
7th Q E: Senti una cosa. Quindi del libro 'Le facce del parlare' ti sono + ti è
+ piaciuto molto il capitolo sui segnali discorsivi?
S: Si.
E: A cosa si riferisce con 'segnali discorsivi'?
Coherence Negotiation in Context 193
2nd Q E: OK. + + Mmhhh. mmhh. Listen, will you talk a little bit abo::ut (05)
second language learning theories + or: at least, not mother tongue.
3rd Q E: This was an approach, a school. And later on? + + Don't you
remember something more recent?
4th Q E: Listen. Let's leave it at that + the hypothesis + + You were talking
instead about descriptive hypotheses. Mmhh? And you mentioned
Austin. Who is Austin?
6th Q E: Okay, ehhh + is there apart of the programme you liked the least?
(05) Which is the part of the programme you liked the least? (05) No,
I won't
ask you about that. (laughs)
7th Q E: Listen. So in 'Le facce del parlare' you liked + you liked the chapter
on 'discourse signals' very much?
S: Yes.
E: What does the expression 'discourse signals' refer to?'
express her opinions and to pose questions to the examiner as well. The
outcome is a much more argumentative and coordinated kind of encounter.
From the point of view of topic, the treatment type of examination tends
to concentrate on fewer, broader and more related topics. One topic leads to
another by contiguity or similarity. The examiner links together the topics under
discussion by way of analogy, comparison, expansion, contrast, etc., much the
way people do in face-to-face conversation. When topic change occurs, it is
normally explicitly declared, justified or explained. The following is an example
of the second type of examination. It is 35 minutes long and organised around
four interrelated questions.
'1st Q E: Now let's take our Orlando an: + well. You can start from a moment,
a point, a Canto, an episode you particularly liked.
2nd Q E: Listen. Would you mind going to Canto 23 where there is another
variation + since we're talking about this poor Petrarca who is used
and changed + let's see i:f + yes, because in the variation we were
talking about before (...) We were talking about changed petrarchism.
4th Q E: Let's change + let's change topic since this one (4 syll) is causing you
problems'
'1st Q: E: Would you mind looking at + the second Canto, you can look at your
notes, outline, or anything you like, and tell me how it is constructed,
what episodes there are, and why it is constructed this way. In short,
+ anything that may come into your mind in order to explain this
Canto of the Poem to somebody else.
196 Anna Ciliberti
[••■J.
2nd Q: E: Well, exactly this - exactly in this second Canto, I think, is the
variation between the two, between Rinaldo and Bradamante.
3rd Q: E: Well, let's go somewhere else. Here + I'll have you stay more or less
on the same topic (...) But then, this declaration in Canto 32 is
comparable to the one in the second Canto?
4th Q: E: Well, listen, I'll have you read the last octave and then you can go,
ok? + I see that you have things very clear. + + Listen, here I - the
topic is quite evident, but I'm not interested in having a thematic
framing in (...)I'd like really a literal explanation of the first octave of
the third Canto.'
Again, the last question has no thematic ties with the previous ones. Yet
the new topic is legitimated by the declared wish to contrast it with the previous
ones, and these are all resumed under a superordinate topic: 'qui io ... non mi
importa tanto il tema inquadrato' ('here, I am not so much interested in the
thematic framing').
An extreme example of a 'treatment' examination is one in which the
student is approximately 35, sophisticated and unusually well-read. What is
striking about this exam is its conversational character. The examination
becomes a symmetrical event in which the student is on a par with the examiner.
Matching assessments and matching stories contribute to the attainment of
coherence both on a local and on a more global level. The whole examination
deals with one single topic: dead languages and how they can signify. In
example (6), the examiner is commenting on a Latin quotation in the candidate's
written assignment:
'E:Why a dead language? And what is the fascination of a dead language? Rather,
the use of a dead language is a kind of necrophilia? (laughs)
S: God + if I could answer this question, ehh, I have to quote one film, I have to
Coherence Negotiation in Context 197
quote 'In the name of the rose', that says, in fact, if I knew I would teach
theology at the Sorbonne.
E: (laughs) Nowadays those who teach at the Sorbonne say if I knew it I would
teach in the United States because they pay more.
(both laugh)'
4 Negotiating coherence
Activity type influences not only the establishment of topical coherence patterns
but also the way coherence is negotiated. People belonging to a common
universe of discourse and sharing the same social goals will try to make their
contributions coherent and will go to any length to detect coherence in what
they hear. This is not necessarily the case when the universe of discourse is not
a common one - i.e. when interactants do not share the same social goals - or do
not attune to one another's individual goals. In the latter case the interaction
may turn out to be non-cooperative and discursively incoherent.
In the exam situation, the student's goal, 'to get through the examination',
is re-defined by the examiner's as 'applying for eligibility to a 'pass". These two
goals may at times (at some level at least) be mutually exclusive and this "may
account for discrepancies in the participants' verbal behaviour" (Sarangi and
Slembrouck 1992: 128). These 'discrepancies' often result in incoherent
sequences.
In the exam situation, in order to reach their goals, students may
consciously - i.e. strategically - produce inconsistent, incoherent answers to the
examiner's questions.
When students do not know the answer to a question, they may take one
of the following two paths: (i) they may openly declare they don't know the
answer; (ii) they may give an evasive answer. Evading a question, or getting
around it, is a typical incoherent way of answering.3
According to Bull and Mayer (1988) question evasion involves the
following strategies:
• apologising
• stating that the question being asked has already been answered
• declining to answer the question
• repeating an answer to a previous question and making a political point
Among this set of strategies, only 'questioning the question' - in the sense
of asking for clarification - is adopted by students during an examination.
Students cannot risk employing other strategies such as 'attacking the question',
or 'repeating an answer to a previous question and making a political point', or
'stating that the question being asked has already been answered'. The most
feasible course of action left to the student would appear to be the adoption of
yet two other strategies:
'E: No. + + Let's see + what is Bahr's manifesto and which are these works?
+ + at least those by Schnitzler and Hofmannsthal.
S:→ We:ll. + + (laughs) My head is completely empty. + + As far as + eh + the
literary works of this period, that is, fundamentally the overcoming of- of
aestheticism as a + + si:mple + eh eh [3 syll]
E: No, let's say, which is the literary manifesto (...)'
Coherence Negotiation in Context 199
In both examples (7) and (8), the students do not know the answers to the
examiners' questions and decide to make use of evasive strategies. In (7) the
student answers in a totally nonsensical way; in (8) the student expands the
scope of the examiner's question. The advance organiser with metatextual
function "intanto" (Tor one thing'/'first of all') announces an insertion
sequence, thus establishing a "hold" (Levinson 1983: 348) in the production of
the second part of the adjacency pair. In this way it functions as a "coherence-
securing means" (Bublitz 1996). The evasive answer goes unnoticed and the
examiner intervenes to correct the student's side sequence. The original
question has been forgotten (turn 5).
The following is the same conversation as it continues:
(9)
6. E: (...) Non necessariamente. E' un ruolo logico, non psicologico, è un
ruolo logico. E' per questo che la Unguistica offre processi di
formalizzazione, eh?
7. S:→Difatti questa ragazza è inginocchiata, forse sta pregando, forse
Brandano, questo non si sa.
8. E: Si, eh eh, per*ché%?
9. S :→*Comun%que, dimostra di essere pentita della vita passata, quindi -
10.E: Sí, ma aspetta, aspetta. Io + qui + tu stai - questo è il limite della tua
cosa. Ricadi sempre nei contenuti. Ennô. Invece, ecco (...)
'6. E: (...) Not necessarily! It is a logical role, not a psychological one, it's a
logical role. It is for this reason that linguistics offers formalisation
200 Anna Ciliberti
processes, right?
7. S:→In fact this girl is kneeling, perhaps she's praying, perhaps Brandano, we
do not know.
8. E: Yes, eh eh, *why%?
9. S:→*Any%way, she shows she has repented for her previous life, therefore -
10. E: Yes, but wait, wait. I + here + you are - this is your limit. You always go
back to content. This is no good. Instead, here (...)'
5 Conclusion
The values that go together with certain pragmatic principles - cooperativeness,
discursive rights/obligations, and the like - and that are critical in relation to
what counts as coherent in interaction depend on the activity type taking place.
Participants in interaction do not have the same rights and obligations, and not
all forms of verbal interaction are dominated by the interactants' wish to be
cooperative. Most importantly, communicators often do not share the same
communicative and/or social goals. But in order for discourse coherence to
apply, interactants must (i) wish to cooperate (in the sense Grice uses the word
cooperation), (ii) have compatible goals, or mutually accepted different goals4,
(iii) communicate within compatible universes of discourse.
The oral examination is an asymmetrical type of encounter in which one
interactant has the institutional role of evaluating the other interactant's
possession of a certain body of knowledge in a certain field. As an activity type,
though, it may serve as a site for multiple and mixed social relations. Various
types of local contexts are intertwined and act upon each other. The examiner
may assume different role identities, thus determining differing levels of
discourse: institutional, professional, interindividual. Coherence - as I tried to
show - will apply differently according to the position - or to the dominant
position - from which the examiner speaks. In what I have referred to as the
'diagnostic' type of examination, the examiner's predominant role is
institutional, i.e. evaluative. She wants the student to report upon a given topic,
Coherence Negotiation in Context 201
(10)
E: No, no La volevo trattenere di più, perché Lei ha lavorato bene + +ma anche
perché, no? + mi pareva di chiaccherare con Lei volentieri (...)
'E: No, no, I wanted to keep you longer, because you have worked well but also I
was enjoying chatting with you'
Notes
1. The examinations were collected by myself and a colleague of mine, Laurie Anderson
2. Transcription conventions
E examiner
S student
+ short pause
++ longer pause
(n) long pause (approximate duration n second)
202 Anna Ciliberti
*text% or
**text%% spoken in overlap with next/previous *text% or **text%%
= latched to previous turn
text - tone group interrupted
text- syllable cut short
text: syllable lengthened (number of colons indicates extent)
.,?! punctuation gives a rough guide to intonation
(?text) text unclear
(?nsyll) tape untranscribable: n = approximate length in syllables
(comment) non-verbal behaviour or context information)
3. See Wilson (1990) for a discussion on the controversial issue of what is and what is not
an answer to a question and, consequently, what counts as an evasive answer.
4. As Sarangi and Slembrouck (1992: 125) note, social equality is what we strive for, not
what we encounter in the social world. But social inequality does not exclude the
possibility to relate with our interlocutors on an equal basis and therefore to produce
coherent discourse.
References
Bergvall, V. and Remlinger, K. 1996. "Reproduction, resistance and gender in
educational discourse: the role of critical discourse analysis". Discourse and
Society 7: 453-479.
Bernstein, B. 1990. The Structuring of Pedagogic Discourse: Class, Codes and
Control, Vol VI. London: Routledge and Kegan Paul.
Bublitz, W. 1989. 'Topical coherence in spoken discourse". Studia Anglica
Posnaniensia XXII: 31-51.
Bublitz, W. 1996. "'I bought a freezer - you've bought a freezer - they've bought a
freezer': Repetition as a text-building device". In C. Bazzanella (ed), Repetition
in Dialogue. Tübingen: Niemeyer, 16-28.
Bull, P. and Mayer, K. 1988. "How Margaret Thatcher and Neil Kinnock avoid
answering questions in political interviews". Paper presented to the British
Psychological Association. London. (Quoted in Obeng 1997).
Chevallard, Y. 1985. La Transposition Didactique: du Savoir Savant au Savoir
Enseigné. Grenoble: La Pensée Sauvage.
Christie, F. 1995. "Pedagogic discourse in the primary school". Linguistics and
Education 7: 221-242.
Ciliberti, A. 1998. "Egocentric vs. self-denying communication in an asymmetric
event". In S. Cmeirkovà et al. (eds), Dialogue Analysis VI. Tübingen: Niemeyer,
29-38.
Coates, J. 1995. "The negotiation of coherence in face-to-face interaction: Some
examples from the extreme bounds". In M. Gernsbacher and T. Givón (eds),
Coherence in Spontaneous Text. Amsterdam: Benjamins, 41-58.
Coupland, N. and Coupland, J. 1997. "Bodies, beaches and burn-times:
Coherence Negotiation in Context 203
1 Introduction
2 Definitions
To begin with, we want to propose a clear distinction between text and
discourse, and correspondingly between cohesion and coherence. These four
concepts crop up continually in the literature, but their relationship generally
remains uncertain and ambiguous. Thus in the work of Michael Halliday, the
terms text and discourse are used in more or less free variation (see Halliday
1992). Wallace Chafe conflates them, too. In his entry in the Oxford
International Dictionary of Linguistics, he has this to say:
The term discourse is used in somewhat different ways by different scholars, but
underlying the differences is a common concern for language beyond the boundaries
of isolated sentences. The term TEXT is used in similar ways. Both terms may refer
to a unit of language larger than the sentence: one may speak of a 'discourse' or a
'text'.' (Chafe 1992: 356)
Both terms may indeed be used to refer to units larger than a sentence,
and many scholars have spoken of either text or discourse as if they were
206 Barbara Seidlhofer and Henry Widdowson
Text
This view of text and discourse is, of course, entirely at variance with that
which holds that text (or equivalently discourse) is a quantitative unit: language
beyond the sentence. Text, in our view, has nothing to do with linguistic extent,
and indeed nothing to do with sentences. Texts come in all shapes and sizes.
They may indeed take the linguistic form of sentences in combination, but that is
incidental. They may equally take the form of isolated sentences (KEEP OFF
THE GRASS) or phrases (WET PAINT) or single words (PRIVATE) or even
letters (P). All such public notices are texts and when we recognise them as
such we engage our contextual knowledge to derive discourses from them, and
read into them what we assume to be the intended reference and force (for
further discussion see Widdowson 1995).
When texts do take the form of larger linguistic units, we can, of course,
consider what links the parts together, and talk about cohesion. Cohesion, then,
is a textual property and has to do with how linguistic elements relate by virtue
of their lexical or grammatical features. But again the occurrence of cohesive
devices are data which have to be interpreted as discourse evidence before you
can make coherent sense of them. This means that they have to be referred to
context. Cohesion is simply the textualisation of those contextual connections
which it is assumed need to be made explicit, but coherence is the discourse
function of realising those connections. It follows that you might derive a
coherent discourse from a text with no cohesion in it at all. Equally, of course,
textual cohesion provides no guarantee of discourse coherence. A simple
example. If you find a plural noun phrase in a text (say, autumn leaves), then
the very copying of the feature of plurality in the following pronoun they
constitutes a cohesive link and this might be read as evidence that the two are
related to a common referent. But consider text (1):
208 Barbara Seidlhofer and Henry Widdowson
(1)
Unfortunately in the days to come autumn leaves will become a hazard to the elderly
as they fall and become a wet soggy mess on the pavement.
and activate it by creating their own discourse from it. They identify the text as
soon as they assign it intentionality. They interpret it as discourse only when
they relate it to the context of their own reality outside language.
text. The aim was to explore how respondents would derive a coherent
discourse from the original text, and how they would regulate their co-operation
according to different instructions. The respondents were Austrian university
students majoring in English, but the fact that they were L2 learners is not of
central importance here, since the educational implications of our observations
apply to all language teaching, and in fact probably to teaching and learning in
any subject.
A. SUMMARY INSTRUCTION
Please write a summary (in no more than 60 words) of the following text, capturing
as faithfully as you can the main points of the author's intended meaning.
B. ACCOUNT INSTRUCTION
Please give a brief account (in no more than 60 words) of what strikes you
personally as of particular interest in the following article. Give your account a title.
The summary instruction thus asked respondents to keep close to the text,
it did not invite them to relate the text to their own reality: they were to make it
coherent on the writer's terms. The account instruction, on the other hand, did
invite the students to engage with the text and make sense of it on their own
terms. These students, then, accessed whatever they found relevant to their own
world. With these different kinds of tasks a mapping-out of the whole range of
responses was targeted, from 'submissive' to author's intention to 'assertive' of
own conditions of relevance. And it was exactly this range of responses that
emerged. To illustrate what the responses were like, here are a few examples:
Coherence in Summary 213
(1)
In the 195 O's, 9% of women of childbearing age had no children; now 25% remain
childless, mostly well educated women from urban areas, who married late and work
outside the home. They either deliberately decide not to have children or postpone it,
preferring their freedom. Later many regret their childless state and try to satisfy
their nurturing instincts at others.
(2)
Those Americans who choose not to have children usually come from urban areas,
are well-educated and marry late. Some of them make the choice deliberately and
some postpone the decision until nature decides for them.
The childless often satisfy their nurturing instincts with nieces and nephews.
Some childless women think they have violated a biological law but most of
them enjoy their freedom.
(3)
In today's America, childlessness is spreading. The childless tend to be well
educated, live in urban areas, marry late and work outside the home. Basically, there
are two groups: the deliberate types and the postponers.
Nonetheless, babies seem to regain their important role. The birth rate among
college-educated women 20 to 24 years old is beginning to climb.
opinions express the current glorification of the individual and the drive to make
changes at will.
It is interesting to note the desire for vicarious parenting, gratification of the
need for human relationships without total commitment.
(8) Statement:
Some people want to have children, others don't. As for the latter, they may
have good reasons for their choice, or they may have none: some just will not spend
part of their lives on raising children, others turn out to regret their hesitating later
on. Personally, I don't care about people's having children or not, and I won't care
about their justification either.
Coherence in Summary 215
8 Participant roles
This is also what Goffman (1981: 144f) seems to have in mind when he talks
about the "production format of an utterance", and there are distinctions
introduced by him which help conceptualise these different roles quite
powerfully. Goffman points out that a speaker/writer can fulfil three different
kinds of role: the animator is somebody who lends his or her voice to the
expression of somebody else's ideas, acting as a "sounding box", as Goffman
puts it. The one responsible for the actual wording of the text is the author,
"someone who has selected the sentiments that are being expressed and the
words in which they are encoded" (Goffman 1981: 144).
Behind these two, however, there is the principal, "someone whose
position is established by the words that are spoken, someone whose beliefs
have been told, someone who is committed to what the words say" (ibid.).
Schiffrin rephrases the distinction as follows: "an animator produces talk, an
author creates talk, [...] and a principal is responsible for talk" (Schiffrin 1994:
104). Schiffrin also points out that these positions can be filled by different
people, but also by a single individual.
Goffman goes on to say that "[t]he notions of animator, author, and
principal, taken together, can be said to tell us about the 'production format' of
an utterance" (op.cit.: 145; see also Couper-Kuhlen, this volume). As
mentioned above, then, Goffman talks about these roles in terms of producers
of language, speakers or writers. But as reading can be seen as creating
216 Barbara Seidlhofer and Henry Widdowson
coherence for oneself, these distinctions are just as useful for thinking about the
reception side of communication. They also tell us about the 'reception format'
of an utterance.
Of course, as is illustrated by the examples from students' protocols
above, many accounts went far beyond an interpretation of the text: not only did
they make the input text their own and express their personal reaction to it, they
actually made the entire communicative event their own and, to use Goffman's
words, "staked out [their] own position" in it - in short, they acted as
principals.
It is important to remember that the distinction between participant roles
is not clear-cut any more than is that between summaries and accounts. It is
actually impossible, and nonsensical, to envisage a summary which does not
involve selection, which after all is something an author, and not an animator
would do. Likewise, it is hard to imagine an absolutely pure instance of
'principal account', without any 'contamination' whatsoever by ideas which are
not entirely the account writer's own. In fact, it would seem that the very nature
of the task, which is intertextual in the case of summaries as well as in the case
of accounts, necessitates that some ideas from the original text be embodied in
the response, whether they become manifest as overt textual features on the
surface or not.
It is clear, then, that the three roles interrelate in complex ways. Just as it
is hard to imagine animating without some degree of authoring, so it is equally
hard to imagine authoring without some expression of attitude, point of view,
etc., i.e. the principal role. In most normal communication, all three are
implicated. The question is the degree of involvement on different
communicative occasions.
For summaries and accounts, we might represent the continuum like this:
and gave their reactions to that while ignoring a host of other issues. What they
got out of the input text was very partial.
But our point is that meaning is partial by definition, and that in language
teaching, and especially foreign language teaching, teachers often create
bafflement by requiring students to derive more complete meaning from a text
than is actually intended to be derived from it. And this is, furthermore, totally
at variance with people's first language experience.
So when talking about coherence and appropriation of meaning we should
think not only about how we can maximise access in the sense of animating, but
also about how we can maximise learners' flexibility and room for manoeuvre,
how we can foster in our students an understanding of the limits of animating.
That is to say that our students need to understand that it is in the nature of
discourse, and of the way people achieve relevance, that meaning is intrinsically
partial and imprecise, that texts are necessarily indeterminate, and that all
readings are approximate. One of the educationally most valuable things we can
do for our students is to make them aware of, and help them cope with, this
necessary indeterminacy and insecurity. Foreign language teachers in particular
tend to be very language-fixated, or code-fixated, and to a certain degree this
has to be so, for otherwise no learning of new language would take place. But it
is too common in foreign language teaching to expect learners to 'understand
everything', and to regard this 'complete understanding' as the most desirable
outcome of any reading or listening activity. This is to say that reformulation in the
form of renderings of a source text very much tends to veer to the 'summary' side on the
left of our diagram. But it is important for learners to explore the whole spectrum, and to
include the other extreme, where they can recast any text in, and on, their own terms. In
other words, teachers would do well to appreciate the fact that coherence cannot be a
fixed, absolute objective, but is always relative to purpose, and we access meaning to the
extent that it is relevant to our purpose. So really any act of comprehension, and any
learning in general, takes place as a personal account. In this sense, then, reformulation is
a very useful activity for enabling learners to make texts coherent for themselves.
References
Chafe, W. 1992. "Discourse: an overview". In W. Bright (ed), International
Encyclopedia of Linguistics, Vol. 1. New York: Oxford University Press, 355-
358.
Goffman, E. 1981. Forms of Talk. Oxford: Blackwell.
Halliday, M.A.K. 21992. An Introduction to Functional Grammar. London: Edward
Arnold.
Schiffrin, D. 1994. Approaches to Discourse. Oxford: Blackwell.
Coherence in Summary 219
1 Introduction
At first sight hypertext does not look like a good subject for research on
coherence. Hypertext is non-linear text, and coherence is typically defined for
linear text. So coherence does not seem to be involved in hypertext at all. But
on closer inspection it emerges that some of the basic structural problems with
hypertexts are classical problems of coherence.
My central question for this paper is: What does hypertext show us about
coherence? But of course the direction of enquiry could easily be reversed by
asking: What does the theory of coherence teach us about the properties of
hypertext? It is interesting to see that until recently much of the literature on
hypertext has been written by computer specialists, specialists in technical
documentation and educational psychologists (e.g. Shneiderman 1992; Kuhlen
1991; Horton 1990; Hofmann and Simon 1995; contributions to Jonassen and
Mandl 1990). Many of these authors have little contact with pragmatics and
text linguistics. As a consequence, some of the aspects which are central to a
pragmatic view of language use still seem to be under-represented in research
on hypertext. It might therefore be useful to see what the hypertext world looks
like from the vantage point of a pragmatic theory of texts. The following
remarks are intended to take a few steps in this direction.
For my theoretical framework I shall assume an action-theoretic concept
of text and dialogue (cf. Fritz 1982).1 In this framework coherence is regarded
as a guiding principle for text production and as the basis for understanding
texts. Understanding a text consists in seeing the relevant internal and external
connections of textual elements. Interpreting a text consists in searching for
and pointing out its relevant connections. According to this theory, coherence
is based on the interaction of different organising principles of texts or
dialogues. In the prototypical case, authors and readers make use of the whole
bundle of organising principles to produce a use or a reading of a text which
has strong coherence properties. Such a reading is functionally coherent,
topically coherent, it is coherent in its knowledge management, and its
222 Gerd Fritz
2 What is hypertext?
The difference between text and hypertext mainly concerns the following
structural and pragmatic properties: A text is a sequence of textual elements T1
... Tn (i.e. sentences or sequences of sentences) which can be used to perform a
sequence of linguistic acts, including the development of a topic. From the
point of view of its author a text can be represented as a pruned tree, as shown
in diagram (1).
(1)
At every node the author has to make up his mind which textual element
to position at this particular decision point. From the point of view of the
reader the textual elements simply form a fixed sequence.
Hypertext, on the other hand, is basically a network of textual elements,
sometimes with hierarchical structures built in. These elements are usually
called topics and the connections provided between the topics are called links.
Diagram (2) shows the structure of a hypertext consisting of the textual
elements T1 ... T8 connected by various links.
Generally, a reader may go through the network in any direction he or
she chooses. For the user hypertext is therefore not really non-linear but multi
linear. A sequence produced by travelling through such a network is called a
path. In the context of this paper it is paths that I am particularly interested in,
for a path is something like a text and therefore it can be evaluated for its
coherence properties.
Coherence in Hypertext 223
3 Paths in hypertext
In discussing the construction of paths through a network I shall mainly
address the following question: How do users of a hypertext make sense of the
path they are following? Generally speaking, within the chosen framework the
answer to this question is: Users make sense of a path or a segment of a path by
seeing sequences of textual elements as realisations of sequencing patterns and
by drawing inferences on the basis of their local and general knowledge. But I
should like to be somewhat more specific than that.
Basically, paths are produced in two different ways, either as pre-defined
paths, which the author presents to the user, or as self-selected paths, which the
user himself chooses from different options available at the individual nodes
within the chosen network. In a way, self-selected paths are the real raison
d'être for hypertext as an interactive medium. Here the responsibility for
making sense of paths is largely shifted to the user. The fact that the user
224 Gerd Fritz
(3)
Word Help Contents
To learn how to use Help, press F1.
Pre-defined paths are closely related to normal linear text. Therefore their
conditions of coherence are quite similar to those of ordinary text. I shall give
two examples, both from the Windows 3.1 online help system, the first one
quite successful, the second one much less so.
Example 1: If you want to know how to create a table in Winword 6.0,
you go to the index of the help system and click your way through to the
following overview "creating a table".
Coherence in Hypertext 225
(4)
Creating a table
®ExarnplesandDemas
Use tables to organize information and create interesting page layouts with side-by-side
columns of text and graphics. In a table, as on a spreadsheet you work with rows and
columns of calls,
To create a table
1. Position the insertion point where you want to create a table.
2. On the Standard toolbar,Clickthe Insert Table button.
From this element links are provided to various other textual elements.
The links are marked by underlined expressions.2 If you choose "examples and
demos" you move to a sequence of small two-part units, in which a step-by-
step instruction is regularly followed by a demonstration of these steps. You go
from instruction to demonstration by clicking on the button "next" and from
there you go to the next instruction by clicking again on "next" and so on.
What we have here is a functional sequence of the basic type: overview
followed by instruction followed by demonstration. This is a type of functional
sequence with which we are familiar from all kinds of teaching, for example in
sports, but also from paper versions of technical documentation. So in this case
the global coherence is grounded on a fiinctional sequencing pattern and on the
continuity of topic.
At this point I should like to digress to briefly discuss an interesting
minor type of sequence which is well known in dialogue analysis. In (4) you
find the word cells underlined in the fourth line. If, as a user, you happen not to
know what a cell is you can click on the word cells and a little pop-up window
226 Gerd Fritz
will tell you: "A cell is the basic unit of a table. In a table, the intersection of a
row and a column forms one cell". If you already knew what rows and columns
are, you now know what a cell is, if not, you will have to move back to the
index and get the necessary information there. From the point of view of
coherence these small explanatory elements are very similar to footnotes or
parenthetical remarks in written text or to so-called side sequences in dialogue
(cf. Jefferson 1972). Sequences of this type interrupt the ongoing dialogue for a
clarification request, followed by a clarification, and lead directly back to the
point of departure. As a kind of question-answer sequence they are themselves
strictly coherent, and as a regular type of insertion they do not disturb the
coherence of the ongoing dialogue either - unless they occur too frequently.
A second type of pre-defined path is a sequence of related topics that can
be accessed by repeatedly clicking the "forward" button. This kind of path is of
course subject to strict conditions of topic coherence. And if anything goes
wrong there - which it easily does - the reader is justifiably upset. My second
example comes from the introduction to Windows Help. This section of
hypertext has a typical hierarchical structure which is, however, not actually
shown to the user. In order to demonstrate what happens to the unsuspecting
user I shall give a reconstruction of the respective hierarchy in the following
tree diagram.
(5)
PROCEDURES
BROWSING MOVING
THROUGH BACK TO CHOOSING
RELATED PREVIOUS HOT-TEXT
TOPICS TOPIC
to know more about "moving in help". You click the button "moving in help"
and this leads you to another button "browsing through related help topics".
This sounds interesting, so you move to this topic. There it says that if you
want to reach the nearest related topic you have to click on the "forward"
button. So that is what you do. And what you get is the topic "Inserting a
footnote" Unfortunately, this is not a closely related topic at all. It is far away
on the next higher rung of the hierarchy and seems to have got into the pre
defined path by mistake. As a novice user you will either think that "Inserting a
footnote" is a strange method of moving in the help system, which is a
misguided hypothesis, or you will suspect that this is a blatant case of topical
incoherence, which in fact it is. It is pleasing to know that this flaw is no longer
found in later versions of the relevant software.
familiar with it, one can explain it to him. Or: By describing this procedure to
someone who has read the description before, one can remind him of the
content of this description. Both patterns are very frequent in instructional
discourse. This kind of structure - explaining by describing - was called level-
generation in Goldman's "Theory of Human Action" (1970) and has been an
important element in other theories of action as well, e.g. in Heringer's
"Practical Semantics" (1978). It is obvious that level-generation plays a very
powerful role in the creation of coherent paths in hypertext. Level-generation
also works for topics. Presupposing appropriate knowledge one can talk about
X by talking about Y, e.g. one can talk about environmental problems by
talking about heating systems.
Due to the hierarchical structuring of many hypertext topics, a very
frequent difference of topic interpretation exists between a top-down or a
bottom-up interpretation. If you move down a hierarchy in a sequence A-B-C,
B will be interpreted as more specific than A, and C as more specific than B.
Alternatively, if you move up the same hierarchy, B will be interpreted as more
general than C, and A as more general than B. The following is a very simple
example which is modelled on structures we frequently find in instructional
hypertexts, e.g. in the teaching programme "Hyperlinguistics" (cf. Ansel and
Jucker 1992; Suter 1995). A short paragraph containing the main aspects of a
theory of grammar (e.g. phonology, morphology, syntax and lexicon) can be
approached from two different directions, i.e. with two different questions in
mind. If we read this paragraph as a partial answer to the question "What
aspects of language does modern linguistics deal with?", its function is to
specify these aspects. If, however, we read the paragraph as an answer to the
question "Where does syntax belong in an overall theory of grammar?", it is
embedded in a different topic. In the first case it belongs to the topic "aspects
of language", in the second case it belongs to the topic "syntax".
All this is theoretically perfectly straightforward. In practice, however,
the user must permanently monitor where he is moving in the hierarchical
structure. And at times this is a very difficult task to perform.
The fact that one can approach textual elements from different directions
also poses interesting problems for knowledge management. In writing good
linear text we try to arrange information in such a fashion that one building
block of knowledge is placed before the next. If understanding block B
presupposes knowledge from block A, we position A in front of B and so on.
Of course, basic conditions of understanding apply to hypertext in the same
way as they apply to linear text. Therefore, as users of hypertext, we must often
be prepared to compensate for the lack of systematic knowledge management.
One of the strategies for this purpose is what David Lewis (1979) called
"accommodation". If, for example, a bit of text begins with the statement "The
Coherence in Hypertext 229
GRAMMAR,
SEMANTICS,
PRAGMATICS
ASPECTS OF A
THEORY OF GRAMMAR
B PHONOLOGY,
MORPHOLOGY,
SYNTAX
SYNTAX
(GRAMMAR)
elements look like hopeful candidates for a useful continuation of his search.
This is somewhat like bringing your interlocutor to make a relevant
contribution to a conversation. Once the chosen chunk of text is presented on
the screen the user has to decide whether it can be incorporated into his
knowledge base as a useful contribution or not. To give an example: If you
search the Internet for information on coherence, you will come across entries
like "The role of coherence in ultrafast chemical reactions" or "coherence
modulating reactive rates" in physics. Maybe your interest in coherence
includes the term coherence and its different uses. In this case you might look
at these topics. If not, you will look somewhere else. You might go to the
homepage of a colleague and find a useful reference to her papers on
"discourse markers". From there you could move on to information on the
"Purdue University On-line Writing Lab". There you might get side-tracked a
bit - but of course you realise that you are being side-tracked - and after a while
you move back to other items on the list. Maybe you will modify your search
topic as you go along and in so doing you will modify your criteria of
coherence. And maybe you will learn new factual connections, and this may
also change your criteria of coherence. If you document your search path you
will probably be able to justify each individual move as a relevant step and
therefore you will classify the whole path as coherent.3 I realise that there are
many open questions at this point. But I shall leave it at that.
5 Conclusions
The main results of my enquiry can be summarised as follows: In hypertext we
get everything from very strong prototypical coherence in guided tours to
minimal coherence in browsing. In self-selected paths forward-looking
coherence construction plays an important role, whereas the role of classical
cohesive ties between textual elements is minimised. As for the concept of
coherence, my observations on hypertext seem to confirm the following
picture: In creating coherence we standardly draw on a whole bundle of
organising principles, but it is possible to deviate from this kind of prototype in
various ways:
(i) It is obviously possible to reduce the amount of cohesive ties like
pronouns, conjunctions and adverbs without losing too much coherence
between textual elements. This loss in explicit marking of coherence relations
is compensated by implicit factors, i.e. by the reader's knowledge of standard
sequencing patterns like functional sequencing, topical progression and so on.
Where this is not the case, lack of explicit marking will often be made up for
by means of inferences.
Coherence in Hypertext 231
Notes
1. Recent developments of this framework can be found in Fritz (1991), (1994), (1997).
2. Names of links are important cohesive elements in hypertext. As opposed to most
cohesive ties in linear text, they are forward-looking.
3. For the connection between the concepts of relevance and coherence cf. Carlson (1983:
45f.) and Hintikka (1986).
References
Ansel, B. and Jucker, A.H. 1992. "Learning linguistics with computers: hypertext as a
key to linguistic networks". Literary and Linguistic Computing 7: 124-131.
Carlson, L. 1983. Dialogue Games, An Approach to Discourse Analysis. Dordrecht:
Reidel.
Fritz, G. 1982. Kohärenz. Grundfragen der Linguistischen Kommunikationsanalyse.
Tübingen: Narr.
Fritz, G. 1991. "Comprehensibility and the basic structures of dialogue". In S. Stati, E.
Weigand and F. Hundsnurscher (eds), Dialoganalyse III, Vol. 1. Tübingen:
Niemeyer, 3-24.
232 Gerd Fritz
(1) The winter of 1963 was very cold. Many barn owls died.
Coherence is that which makes a discourse more than the sum of the
interpretations of the individual utterances. A set of sentences is coherent if and
only if all of the segments in the discourse structure are connected to each other
(cf. Mann and Thompson 1988). Whatever the merits of this definition, it makes
clear that coherence is a notion that applies to the level of the discourse
representation.
How can we account for the coherence of discourse? This is an important
question in both linguistic and cognitive studies of discourse. At least two
different leading approaches can be distinguished in recent research on
coherence, the relational approach and the intentional one.3 In the first
approach, represented prominently by Hobbs (1979) and Mann and Thompson
(1988), coherence is modelled in terms of rhetorical or coherence relations like
Result and Concession that exist between discourse segments (the basic
elements of connected discourse, which correspond minimally to clauses).
236 Ted Sanders and Wilbert Spooren
(2) I am sure it was a sparrow-hawk. It had a grey back and brown stripes
on its chest.
The crucial point in Moore and Pollack's proposal is their claim that a complete
computational model of discourse structure cannot depend upon analyses in
which the informational and intentional levels of relation are in competition. [...]
In RST, and, indeed in any viable theory of discourse structure, analyses at the
informational and the intentional level must coexist. (1992: 538)
This claim can be taken to imply that each time a rhetorical relation can be
identified between two consecutive discourse segments, there are in fact two
relations, or rather two levels at which the relation can hold, the intentional and
the informational level. Although it is not clear whether Moore and Pollack
would agree with this position (they formulate a somewhat different view
elsewhere in the very same paper)4, we want to explore its implications for
discourse structure theory. Let us assume this position to be the strong version
of the multi-level thesis. It incorporates the following two claims:
a. two relations hold simultaneously between two discourse segments (an
informational one and an intentional one)
b. the speaker's intentions help to account for the coherence of discourse.
Moore and Pollack's argument for the claim that informational relations
and intentional relations co-exist simultaneously rests on the following example.
(4)
a George Bush supports big business.
b He's sure to veto House Bill 1711.
(5)
De weg was geblokkeerd doordat er een lawine
The road was blocked because there an avalanche
was geweest op Roger's Pass.
was been on Roger's Pass.
(6)
a S1: I want to lift the piano.
b S2: ok.
c I will pick up this end.
One of the things Grosz and Sidner argue is that both participants infer
something like (7) from segment (6)c:
Coherence Relations 239
(7)
MB(S1,S2,INT(Si,lift(foot-end)))
(where MB=Mutual Belief, INT=Intention)
But that does not tell us how such an intention can account for the
coherence of the discourse representation of the fragment, i.e. in what way the
segments are connected to each other.
Should it come as a surprise that it is so difficult to align the concepts
'intention' and 'coherence relation'? We believe the answer is no, because
intentions and coherence relations have a very different ontological status.
This is our third argument against the strong multi-level thesis: intentions
and relations are different types of entities (see Asher and Lascarides 1994 for
additional reasons)6. Intentions are basically unary in that they are functions, the
arguments of which are the speaker, the hearer and a single proposition. By
contrast, coherence relations are minimally binary, in that they connect at least
two propositions. A typical discourse intention is "the speaker wants to
persuade the hearer that a certain state of affairs is the case/worth persuing/..."
(WANT(S,(BELIEVE(H,P)))). This is ontologically very different from
connecting two states of affairs or events in, for instance, a Cause-relation
CAUSE(P,Q).
The approach advocated here is typically relational in the sense that the
coherence of discourse is accounted for in terms of coherence relations that
exist between segments. These relations are conceptual by nature because they
are not part of the discourse, but of the representation language users have or
make of a discourse. A number of theories have made use of relations in
explaining coherence (Hobbs 1990b; Longacre 1983; Mann and Thompson
1986, 1988; Martin 1992; Meyer 1975). Yet, there is no consensus about a
single set of relations (see, for instance, Hovy 1990), and the alternative sets
that have been put forward are very different. Furthermore, many sets of
relations are presented as plain lists, unorganised and extendable ad infinitum.
240 Ted Sanders and Wilbert Spooren
This is unsatisfactory for several reasons. The first and most important
one is that in this situation it is difficult to account for the way in which a
coherent cognitive representation can be constructed. For instance, how does a
reader arrive at the interpretation of a particular coherence relation such as
Evidence? If all relations are considered to be on the same basic level, it must be
assumed that readers use their knowledge of all these relations (30, 100, 1000?),
in order to interpret a stretch of discourse; it must further be assumed that
Evidence, and Solutionhood, and Frustrated Expectation as well as many other
relations which scholars have identified, but about which they do not agree, are
all cognitively basic. A second reason to feel discomfort about such lists is that
students of coherence relations seem to agree about some basic notions, which
occur in every proposal. Examples are the causal and contrastive nature of
relations. A third point to consider is that the way in which coherence relations
are expressed linguistically reflects the relationship that exists between the
different coherence relations. There are restrictions on the type of relation a
connective can express. For example, and can express additive and causal
relations but not concessive relations. And however can express contrastive and
concessive relations, but not causal relations (see, for instance, Degand 1996;
Knott and Dale 1994; Knott and Sanders 1998 and Pander Maat 1998 for some
systematic studies). These restrictions imply an organisation of the relations that
connectives can express, and an adequate theory of discourse structure should
be able to express these restrictions.
Hence, an adequate theory of discourse structure will have to explain the
fact that the similarity between coherence relations varies. To that end, the set
of relations should be categorised, and it can then be regarded as a list which is
in principle finite.
In Sanders, Spooren and Noordman (1992, 1993) we have proposed a
classification scheme which accounts for the 'relations among relations'. The set
of coherence relations is described in terms of four primitives, shared by all
relations: Basic operation (relations are causal or additive), Polarity (positive or
negative), Source of coherence (semantic or pragmatic) and Order of segments
(basic or non-basic order). In these terms, the examples (1) and (3) would be
categorised as causal, positive, semantic, basic order {Cause-Consequence) and
sequence (2) would end up as causal, positive, pragmatic and non-basic order
{Claim-Argument).
The four primitives are important cognitive categories, prominent in
research on language and language behaviour. Predictions based on this
relational classification theory have been tested in several experiments. In
Sanders et al. (1992) it was shown that the coherence relations that are similar
in terms of these primitives were confused more often than relations that are less
Coherence Relations 241
In (8) but not in (9) the second clause can be taken to be a cause for the
effect in the first clause. This Cause relation is a relation between states of
affairs. In (9) there is not such a relation between states of affairs. Here the
second clause gives a justification for uttering the first clause. Following the
terminology of van Dijk (1977) we use the terms semantic and pragmatic
relations to describe the difference. In view of the rich and laden history of these
phrases it is perhaps wiser to use more neutral terminology: 'propositional' for
relations between states of affairs and 'illocutionary' for relations between
speech acts. 7 All relations that are called intentional in the literature, for
instance by Moore and Pollack (1992), are of the illocutionary type, whereas
the informational ones are propositional.
The relevance of a distinction between propositional and illocutionary
relations comes from the fact that several researchers have suggested that
connectives exist across languages, which can be considered 'domain-specific',
i.e. they can only be used to express illocutionary relations or only to express
propositional relations. This has been suggested for Japanese kara versus node
(Takahara 1990), for French car versus puisque (Bentolila 1986) and for
German denn versus weil (e.g. Günthner 1993; Keller 1995); Németh (1995)
suggests that Hungarian hat and mert are pragmatic connectives, Sweetser
(1990) suggests that English since prefers non-propositional relations, and
Knott (1996) and Knott and Sanders (1998) show that in both English and
Dutch domain-specific cue phrases exist.8
In sum, we agree with Moore and Pollack (1992) that coherence relations
come in two sorts, but we also think that this property of relations should not be
confused with the entirely different notion of intention.
obvious that any account of discourse must deal somehow with communicative
intentions and their link to discourse structure. The question is how to account
for this connection.
From a cognitive point of view it seems very well possible to combine the
two concepts, for instance on the basis of one of the leading theories of
discourse production, namely Levelt's (1989) model of speech production. In
this model there is a planning component, the Conceptualiser, in which
communicative intentions are mapped on speech acts, which in turn are mapped
on the preverbal message. The latter is input to the subsequent components of
the model, such as the Formulator, in which lexical items are selected and
syntactic structures are built. In this model the preverbal message is the
(preverbal) realisation of the communicative intentions, and we suggest that
coherence relations should be located exactly there: They are a means of
realising complex communicative intentions.
(10) List
a. George Bush is the 53rd president of the US.
b. He's from Texas.
c. He is bound to sign the economical treaty with Canada this month.
d. He supports big business.
e. He's sure to veto House Bill 1711.
f. He has a keen interest in foreign politics.
g. And he is the president of the Washington Rotary club.
Coherence Relations 243
PLANNING COMPONENT
Communicative intentions
DISCOURSE STRUCTURE
a. Semantic information
b. Coherence relations
(i) 'propositional'
(ii) 'illocutionary'
DISCOURSE
(11) Claim-Argument
a. Last week's Newsweek contained an article
b. revealing that behind the screen George Bush is a fervent supporter of
the Texan oil billionaires.
c. This has been suggested previously,
d. but this is the first time that a reporter gave a conclusive argument for
that claim.
e. George Bush supports big business,
f He's sure to veto House Bill 1711.
(12) Enumeration
a. Republican Presidents have always been strongly opposed to the
environmental lobby:
b. George Bush supports big business.
c. He's sure to veto House Bill 1711.
d. On top of that, he has always refused to talk to activist groups like
Greenpeace.
(13) Contrast
a. You never know what to think of this guy Bush.
b. He supports big business.
c. He's sure to veto House Bill 1711.
d. So his politics are totally inconsistent.
The multi-level thesis originates in the observation that in, for instance, a
Claim-Argument relation (as in (11)), the argumentation relies on certain states
of affairs in the world. For instance in (11), House Bill 1711 must have a
character that is detrimental to big business. But we think that it is misleading to
say that in (11) there is a Cause relation between e and f, because the link
between the two segments is argumentative and not causal. Evidence for this
claim is that the Claim-Argument relation can co-vary with a whole range of
states of affairs in the world: time relations (John's the murderer, because
directly after the murder I saw him at the scene of the crime), categorisation
relations (It is a singing bird, because it is a robin), listing (or even contrast)
Coherence Relations 245
relations {It must have been my mother who gave me the tennis balls for my
birthday, because my father gave me the tennis racket), something that one
might want to call source relations (John isn 't coming, because he told me so),
and 'hypothetical world' relationships (You must borrow me ten dollars,
because otherwise I'm not speaking to you anymore).
More generally, the fact that out of context relations can be vague does
not imply that there is always more than one relation, let alone that when there
is a relation at the informational (propositional) level there is also a relation at
the intentional (illocutionary) level.
The conclusion is that for example (4) there are many more possible
interpretations than just the two identified by Moore and Pollack. Furthermore,
the examples (10)-(14) show that, within context, it is perfectly clear which
relation connects the segments: they can be classified in terms of the primitives
given in Sanders et al. (1992). The relations in examples (10) and (13) exist only
at an ideational/propositional level (as in Moore and Pollack's original
Volitional Cause reading), and in examples (11), (12) and (14) at an
intentional/illocutionary level10 (as in Moore and Pollack's original Evidence
reading).
(15)
Dankzij dat veel te kleine lontje heb ik nu eindelijk een hondje.
'Thanks to that fuse that was much too short, I finally have a doggy for my
support.'
connective thanks to and the adverb finally and that at the same time captures
the advisory nature of the authors' intention. To us it seems obvious that at the
level of discourse coherence, the authors express a Volitional Result relation,
whereas at the level of discourse intentions they try to warn the readers. Hence,
although it is clearly possible to detect an intentional message in the
utterance - for instance, the linguistic form strongly signals 'gratitude' - the
connectedness of the discourse cannot be characterised in terms of its intention.
The reason is that there is a clear mismatch between the linking of the clauses
and the communicative message that the authors try to convey.
There is a more general point to be made here: does connected discourse
cohere because the author wants to convince the reader or does it cohere
because the author gives evidence for a claim? We think the latter is the case.
To convince someone you do not even need words, you could just as well point
a gun at him. And if you use words, why would you need more than a single
proposition? In other words, the level of intentions is not suited to account for
the coherence of a discourse.
In short, our proposal is to keep communicative intentions and coherence
relations apart and to look upon coherence relations as realisations of intentions.
Such a cognitive account predicts close relationships between intentions and
relations: When a writer intends to inform the reader about some event in the
world, (s)he will typically choose propositional relations and when the intention
is persuasive, a writer will often use illocutionary relations. Such a prediction is
corroborated by the finding that there are strong correlations between
coherence relation type and discourse type (cf. text type in Virtanen 1992). In a
corpus of argumentative and descriptive texts it was found that relatively more
illocutionary relations occur in argumentative texts than in descriptive texts.
This relative predominance of illocutionary relations also holds in terms of the
amount of relations at the highest text levels (Sanders 1997). And in an
experimental study on the acquisition of connectives and relations we found that
the coherence relations between utterances produced by primary school children
were strongly determined by the discourse task in which the children were
involved: a picture description task resulted mainly in propositional relations;
when asked for an opinion the children mainly produced illocutionary relations
(Spooren, Tates and Sanders 1996).
The picture that emerges is the following: intentions and discourse
coherence are to be separated; coherence relations are realisations of
communicative intentions. Some intentions are realised in a preverbal message
without any coherence relation, others are realised in a preverbal message
containing an illocutionary relation, and still others are realised as a
propositional relation. For instance, if I were to have the communicative
Coherence Relations 247
intention to convince you that George Bush supports big business I might try to
realise this intention by stating that George Bush supports big business. But I
may also believe that merely uttering this will not do the job. In that case I may
add an argument to my utterance, namely that Bush is sure to veto House Bill
1711 (which is the analysis of example (11); in this reading the bill is detrimental
to big business). Alternatively, I may have the communicative intention to
convince you that Bush's politics are totally inconsistent, and again, I also
believe that merely uttering this will not convince you and therefore I present as
an argument the contrast between Bush's support of big business and his
upcoming veto on House Bill 1711 (example (13); in this reading the bill is
favourable toward big business). Finally, I might want to describe to you the
main characteristics of Bush's political life among which his support of big
business and his veto on House Bill 1711 (example (10)).
Such a view is 'multi-level' in the sense that speakers and hearers keep
track of their (alleged) intentions and the discourse structure simultaneously.
But it does not imply a discourse structure which is multi-level in itself, in that it
consists of an intentional discourse structure separated from and coexisting with
an informational discourse structure.
Notes
1. Authorship of this paper is shared equally. Earlier versions of this paper were presented
at the 4th Conference of the International Cognitive Linguistics Association in
Albuquerque, NM, July 1995 and at an International Workshop on Discourse Markers in
Duisburg, Germany, October 1995. We would like to thank several people for comments
on an earlier draft, especially Ed Hovy and Alistair Knott. Needless to say that all
remaining errors are ours.
2. Strictly speaking, the coherence is not a property of the discourse but of the
representation language users make of the discourse.
3. We leave aside approaches dealing with referential coherence, i.e. repeated reference to
the same set(s) of entities. See Sanders, Spooren, and Noordman (1992), section 1.1 for a
discussion and relevant literature.
4. "We are not claiming that interpretation always depends on the recognition of relations
at both levels, but rather that there are obvious cases where it does. An interpretation
system therefore needs the capability of maintaining both levels of relation" (Moore and
Pollack 1992: 540).
5. Empirical evidence for this claim comes from the finding that when exclusive cases are
presented to judges, they intuitively agree with the classification of items in terms of
'informational' or 'intentional' (or, in the terminology to be introduced later,
propositional and illocutionary, respectively), independently of the context in which the
items are presented (Sanders et. al. 1993; Sanders 1997).
248 Ted Sanders and Wilbert Spooren
6. Asher and Lascarides (1994) outline a formalisation of the relation between intentional
structure and discourse structure using a system of non-monotonic reasoning and argue
for a distinction among the two on different grounds, namely that the rules governing the
two types of structures behave differently. For instance, intentional structure allows for
abductive reasoning, and discourse structure does not.
7. We are aware of the fact that each segment has a locutionary and an illocutionary
meaning (Pander Maat 1994), but that does not imply that the link between discourse
segments is to be located at the locutionary and the illocutionary level. Coherence
relations like Cause connect segments at a propositional level, whereas Justification and
Evidence relations connect segments at an illocutionary level.
8. At the same time, corpus-analytic research shows that the distribution of many common
connectives, for instance Dutch causal conjunctions, cannot be explained in terms of
these 'domains' only, see for instance Pander Maat and Sanders (1995), Pit, Pander Maat
and Sanders (1997).
9. The model suggests that language production proceeds serially. This is only partially
correct, in that the model works incrementally. as soon as a minimal unit is composed by
one component, it is passed on to the next component. In this way the various
components can work in parallel, be it on different parts of the message.
10. This is confirmed in experiments in which language users are asked to judge relations
(Sanders 1997).
References
Andriessen, J., de Smedt, K. and Zock, M. 1996. "Discourse planning: empirical
research and computer models". In T. Dijkstra and K. de Smedt (eds),
Computational Psycholinguistics. London: Taylor and Francis, 247-278.
Asher, N. and Lascarides, A. 1994. "Intentions and information in discourse".
Proceedings of the 32nd Annual Meeting of the Association of Computational
Linguistics June, 27-30, 1994. New Mexico.
Bateman, J.A. and Rondhuis, K.J. 1997. "'Coherence relations': towards a general
specification". Discourse Processes 24: 3-49.
Bentolila, F. 1986. "CAR en français écrit". La Linguistique 22: 95-115.
Bratman, M.E. 1990. "What is intention?" In P.R. Cohen, J. Morgan and M.E. Pollack
(eds), Intentions in Communication. Cambridge, MA: MIT Press, 15-32.
Degand, L. 1996. A Situation-Based Approach to Causation in Dutch with some
Implications for Text Generation. Dissertation, Université Catholique de
Lou vain, Belgium.
Dijk, T. van. 1977. Text and Context. Explorations in the Semantics and Pragmatics
of Discourse. London: Longman.
Gernsbacher, M. and Givón, T. (eds). 1995. Coherence in Spontaneous Text.
Amsterdam: Benjamins.
Grosz, B.J. and Sidner, C.L. 1986. "Attention, intentions and the structure of
discourse". Computational Linguistics 12: 175-204.
Coherence Relations 249
Grosz, B.J. and Sidner, C.L. 1990. "Plans for discourse". In PR. Cohen, J. Morgan
and M.E. Pollack (eds), Intentions in Communication. Cambridge, MA: MIT
Press, 417-444.
Guenthner, S. 1993. "'...weil - man kann es ja wissenschaftlich untersuchen' -
Diskurspragmatische Aspekte der Wortstellung in WEIL-Sätzen". Linguistische
Berichte 143: 37-59.
Hobbs, J. 1979. "Coherence and coreference". Cognitive Science 3: 67-90.
Hobbs, J. 1990. Literature and Cognition. Menlo Park: Center for the Study of
Language and Information.
Hovy, E. 1990. "Parsimonious and profligate approaches to the question of discourse
structure relations". Proceedings of the 5th International Workshop on Natural
Language Generation: 128-136.
Keller, R. 1995. "Das epistemische WEIL". In R. Keller (ed), Zeichentheorie.
Tübingen: Francke, 239-252.
Knott, A. 1995. A Data-Driven Methodology for Motivating a Set of Coherence
Relations. Ph.D. Thesis, University of Edinburgh.
Knott, A. and Dale, R. 1994. "Using linguistic phenomena to motivate a set of
coherence relations". Discourse Processes 18: 35-62.
Knott, A. and Sanders, T. 1998. "The classification of coherence relations and their
linguistic markers: an exploration of two languages". Journal of Pragmatics 30:
135-175.
Kroon, C.H.M. 1995. Discourse Particles in Latin: A Study of NAM, ENIM, AUTEM,
VERO and AT. Amsterdam: J.C. Gieben.
Levelt, W.J.M. 1989. Speaking. From Intention to Articulation. Cambridge, MA: MIT
Press.
Longacre, R.E. 1983. The Grammar of Discourse. New York: Plenum Press.
Mann, W. and Thompson, S.A. 1986. "Relational propositions in discourse".
Discourse Processes 9: 57-90.
Mann, W. and Thompson, S.A. 1988. "Rhetorical structure theory: toward a functional
theory of text organization". Text 8: 243-281.
Martin, J.R. 1992. English Text: Systems and Structure. Amsterdam: Benjamins.
Meyer, B.J.F. 1975. The Organization of Prose and its Effects on Memory.
Amsterdam: North-Holland.
Moore, J.D. and Pollack, M.E. 1992. "A problem for RST: The need for multi-level
discourse analysis". Computational Linguistics 18: 537-544.
Németh T., E. 1995. "On the role of pragmatic connectives in Hungarian spoken
discourse". In B. Wårvik, S.-K. Tanskanen and R. Hiltungen (eds), Organization
in Discourse. Proceedings from the Turku Conference. Turku: University of
Turku, 393-402.
Pander Maat, H. 1994. "Coherentie-relaties hebben zowel semantische als pragmatische
aspecten". [Coherence relations have both semantic and pragmatic aspects]. In A.
Maes, P. van Hauwermeiren and L. van Waes (eds), Perspectieven in
Taalbeheersings onderzoek. Dordrecht: ICG, 118-129,
250 Ted Sanders and Wilbert Spooren
Willis J. Edmondson
Universität Hamburg
between discourse and text - like the distinction between utterance and
sentence - is not one of substance, but one of use. How far in reading a
transcription (or, indeed, viewing a video-recording of a discourse event), we
are in fact creating a discourse distinct from the one that was transcribed and/or
recorded, is, in a sense, precisely the question I wish to raise in this paper.
Let me now briefly gloss the term coherence. To refer to the coherence of
a discourse is to refer to the ways in which its parts constitute a whole. We are
talking about whether and how a discourse makes sense, and this making-sense
is constituted by the behaviours realised or represented in the words, sentences
and texts used in the discourse, and not by these linguistic tokens themselves.
By contrast, the way or ways in which elements in a sentence or in a text fit
together is a matter of grammar and of cohesion. Now if a discourse is a social
event, and if, as is surely the case, a social event only makes sense for members
of the society in which it has a purpose, then it follows that coherence is a
matter of interpretability. The coherence of an instance of language in use is
therefore a reflection of the coherence-making skills of the participants and/or
the researcher. A discourse cannot therefore be said to be coherent without the
mediation of a human mind.
If these rather commonplace views on the nature of discourse and of
coherence are accepted, what, then, can we say of the term discourse analysis
itself? I shall assume that discourse analysis is a principled undertaking,
aspiring to scientific status, and centrally concerned with explicating the nature
of the coherence residing in a discourse. Other, additional goals might also
hold, but they will be premised on the discernment of coherence. If, I wish to
claim, you are not interested in asking how the participants "make sense" of the
piece of language you are handling, then you are not doing discourse analysis,
on my understanding, but some other thing.
Returning to the question raised above concerning the status of the
analyst's interpretation of a transcription or of a video-recording, where, we
may now ask, does then the discourse analyst fit in, in terms of the coherence-
creating capacities of the human mind? Does he or she establish coherence
additional to, or independently of the discourse participants themselves? What
validation criteria hold for a particular analysis of a given piece of data? If
coherence is achieved, as my title rather quaintly puts it, then where doth
meaning lie?
Coherence and Meaning 253
analyses seem to be validated only in terms of the results their analyses lead to
- results which are themselves predictable, given the philosophy behind them.
Thus a resultative stance is taken, in order to support a relativistic view.
The resultative stance can further be seen to operate when the interest of
the data constitutes the interest of its analysis. Consider for example Peräkylä
(1993). The paper is rich in data, but remains, as far as I can see, descriptive.
The major point that emerges from the analysis of counsellor behaviour is as
follows: "As a summary, the recurrent management of the epistemological
framework of the counsellors' description of the clients' future seems to display
the counsellors' orientation to such descriptions as sensitive and delicate
matters". The empirical basis for this conclusion is essentially the observation
that in these counselling sessions, AIDS is mentioned, if at all, only as a
hypothetical possibility: death is mentioned, if at all, as something that will
inevitably confront Everyman. The author suggests finally that the analysis
attains a certain poignancy because behind the hypothetical a real possibility is
lurking: in other words, AIDS and/or death may not be far away.
Voss et al (1996) claim that in a "constructivist" framework, the
assumption is that "the reader or listener constructs a mental representation that
is a fonction of the discourse contents and characteristics of the writer". For
these authors, race and attitudes influence discourse processing (p. 104). This is
an extraordinary claim. What the authors seem to mean is that race and attitude
influence one's perceptions and social values. This is certainly true, but this has
little to do with discourse processing, I venture to suggest.
Consider too Kurzon (1996). The violation of the maxim of quantity in
terms of hyponymy is illustrated via an analysis of a broadcast interview given
by Princess Diana. The sole point of the article seems to be to document the
fact that the interviewee referred often enough in this interview to "'people in
my environment'", instead of saying, perhaps, "the Queen", "Mum", "Liz", or
whatever.
One does not wish to deny the validity, and indeed the inherent interest,
of such data, and of such analytic results. I do however want to cast doubt on
the claim that such analyses are necessary because of their plausibility and
interest instances of discourse analysis in a technical sense.
The relativistic stance - the view that nothing goes because anything goes (or
vice versa) - implies that there are no discourse universals. Therefore the
degree to which we can argue the opposite may determine our success in
demonstrating that the discourse analyses we propound and practice have a
valid grounding outside of our own discourse interpretive skills, intuitions, or
prejudices. A four-levelled model might be proposed, as follows.
Level 1 is biological, and focuses on the deep-seated tension between
individual and social drives. Noli me tangere and Come together! Self versus
Other as opposed to Self via Other. It can be shown that interaction rituals have
been developed to reconcile these drives. At this level, then, some basic
interactional structures can be posited, as they can be discovered in many
animal species, and in other semiotic systems than human language. The tied-
pair of ethnomethodological infamy is the most simplistic example.
Level 2 we may call sociolinguistic. Here the assumption operates that,
with humans language has further been developed and adopted - amongst
other things - to serve and reconcile these biological drives operating at level
1. When language comes into the picture, further interactional structures
become possible, for two reasons. Firstly, linguistic symbols can themselves
function as social tender - thus an apology both stands for and constitutes an
act of redress. Secondly, language itself can be topicalised. We can therefore
talk about talk. This has structural consequences. For example, if I solicit your
bread, and you comply with this request, the transaction is potentially closed:
if, however, I solicit your opinion, and you comply with this request, you
260 Willis J. Edmondson
thereby open an exchange, rather than closing one (for details of such
structures, cf. Edmondson 1981a).
But the main thrust of the argument at level 2 is that certain outcomes
have to be possible via talk if human society is to function. Further, various
'speaker meanings', in my terms, have to be possible, if such outcomes are to be
reached via discourse. For example, agreement on a cooperative undertaking is
surely an outcome without which no social planning, task-sharing or
cooperative achievement would be possible. If such an outcome is to be
reached, and if speech is to play a decisive role to this end, then an utterance
whereby a speaker indicates the desirability of some future collective action is
necessary, as is an utterance whereby a speaker indicates his or her willingness
to participate. At this broad sociohnguistic level, the claim is not that things
like suggestions, offers, requests, apologies and so on are universal categories
of "speech acts" or speaker meanings. The claim is simply that such speaker
meanings, which enable discourse outcomes which fulfil biologically-based
social needs, will be of interactional significance in any speech community.
The third level I want to call interactional. At this level very broadly-
based constraints on human interaction operate. Roughly, the idea is that level
1 gives us biological constraints and goals. Level 2 gives us some requisite
discourse outcomes, together with some communicative categories needed in
order to reach them, while level 3 concerns general rules which govern the
exchange of such speaker stances in discourse interaction. It is a question of
general performance constraints. This is then the level at which conversational
maxims, theories of politeness or of relevance operate. It is very broadly-based,
and concerns essentially the ways in which the universally-relevant categories
and/or outcomes at level 2 can be realised linguistically (or indeed by other
communicative means).
The fourth level is that of culturally-institutionalised realisation. It is the
level of empirical observation. Different discourse conventions clearly hold in
different institutionalised contexts inside and outside one culture, whether
national, linguistic or professional. It is all too often at this level that cross-
cultural comparisons are made, and a search for universals is implemented.
General constraints may, of course, be found to operate at level 4 for all
languages, but it is unlikely that they will, except when such constraints are
formulated at an exceedingly high level of generality, i.e. at level 3 above, or
higher. Thus, for example, Kasper (1995) worries about the bases of
contrastive pragmatics; Rose (1994) has claimed that the discourse completion
tests used for various language groups inside CCSARP - a multi-cultural
research project (cf. Blum-Kulka, House and Kasper 1989) - may simply not
apply in some cultures at all, a worry voiced already in Wolfson et al (1989).
Coherence and Meaning 261
At this level, too, of course, changes occur over time: cultural discourse norms
are not static.
Clearly, the above constitutes at best the bare outlines of a multi-levelled
theory. It suggests, though, that relativistic views of discourse underplay the
universal core that underpins all human verbal interaction. Following this
approach, then, specific types of discourse behaviour could be related to
various kinds of discourse universals. It is an interesting fact, I think, that
inside linguistics, the a priori evidence for a universal theory of discourse is
much more present and biologically warranted than the evidence for a
universal theory of grammar, though universal bases for linguistic competence
have been investigated in great detail, while a universal base for discourse or
communicative competence has received very little attention.
6.1 World-switching
Firstly, world-shifting and switching can occur quite rapidly inside one flow of
text - Tristam Shandy is an early literary exploitation of this fact:
My mother, you must know.. but I have fifty things more necessary to let you know
first..I have a hundred difficulties which I have promised to clear up, and a
thousand distresses and domestic misadventures crowding in upon me, and
threefold, one upon the neck of another. A cow broke in (tomorrow morning) to my
uncle Toby's fortifications, and ate up two rations and a half of dried grass.. Trim
insists on being tried by a court-martial- the cow to be shot- Slop to be crucified..
I want swaddling., but there is no time to be lost in exclamations.. I have left my
father lying across his bed and promised I would go back in half an hour; and five-
and-thirty minutes are lapsed already.
(Tristam Shandy, p. 170)
262 Willis J. Edmondson
Secondly, not only do speakers switch frame, as it were, inside one and the
same discourse context, but they may simultaneously address different
audiences at the same time, transmitting different speaker meanings to them via
the same utterance or set of utterances. So the notion of co-existent discourse
world encompasses here superimposed communicative networks. For example,
remarks addressed to small children in the presence of their parents are
commonly primarily addressed to those parents. A teacher may give positive
feedback to an individual pupil, and at the same time tell the rest of the class
that the answer was not quite right.
7 Summary
In this theoretical paper, I have put forward a view of discourse and of
discourse coherence which necessarily invokes the perceptual and interpretive
skills of human minds. I went on to suggest that the nature of discourse raises
problems of validation inside discourse analysis. To resort to relativistic
approaches is however to bypass the issue of validation, and is inconsistent
with viewing discourse analysis as a principled, scientific undertaking. In an
attempt to go some way towards avoiding relativity, but respecting discourse
data, I then presumed to sketch some elements which together constitute a
rudimentary universal theoretical base for discourse undertakings. Finally, I
went on to look at the notion of co-existing frames of reference or "discourse
worlds", as one conceptual attempt to reconcile the implications of the
multiple-meanings hypothesis with the requirements of theoretical and
structural explicitness.
As said at the beginning of the paper, the term "discourse analysis" is
used with very many senses, and I do not wish to claim that the view of
discourse analysis I have taken in this paper is in some sense primary. I do
wish to claim, however, that if discourse analysis is seen as a principled
approach to handling instances of language in use, then it has to be possible to
argue about a given piece of data analysis. This means, as far as I can see, that
the issue of analytic validation cannot be avoided.
Notes
1. The four-volumed Handbook of Discourse Analysis (van Dijk 1985) is still good evidence
for this observation. More recently, Schiffrin (1994) has devoted a book to different
concepts of discourse. On different concepts of coherence, cf. for example Sanders et al
(1992).
2. I want to refer to 'speaker-meaning' as the stance adopted by a speaker at a specific point in
the discourse via a discourse move. More abstractly, I shall refer to discourse meaning to
indicate the significance of any unit of discourse inside the whole. Inside the theory of
discourse that I uphold, speaker meaning and discourse meaning are not the same thing.
The latter is roughly made up of the former in the light of its structural placing. So the
meaning of a discourse as such derives from its constituting a coherently structured
sequence of speaker meanings. Cf. for example Edmondson (1981a) for details.
References
Blum-Kulka S., House, J. and Kasper, G. (eds). 1989. Cross-Cultural Pragmatics:
Requests and Apologies. Norwood: Ablex.
Dijk, T. van. (ed). 1985. Handbook of Discourse Analysis. London: Academic Press.
264 Willis J. Edmondson
Edmondson, W.J. 1981a. Spoken Discourse: A Model for Analysis. London: Longman.
Edmondson, W.J. 1981b. "Illocutionary verbs, illocutionary acts, and conversational
behaviour". In H.-J. Eikmeyer and H. Rieser (eds), Words, Worlds and
Contexts. Berlin: de Gruyter, 485-499.
Edmondson, W.J. 1985. "Discourse worlds in the classroom and in foreign language
learning". Studies in Second Language Acquisition 7: 159-168.
Edmondson, W.J. 1989a. "Principles of cooperation in interaction". In R. Dirven (ed),
A User's Grammar of English: Word, Sentence, Text, Interaction. Frankfurt:
Lang, 809-840.
Edmondson, W.J. 1989b. "Discourse production, routines and language learning". In
B. Kettemann et al. (eds), Englisch als Zweitsprache. Tübingen: Narr, 287-302.
Edmondson, W.J. 1990. "Can one usefully do discourse analysis without investigating
discourse processing?" In H. Nyyssónen et al. (eds), Proceedings from the 2nd
Finnish Seminar on Discourse Analysis. Oulu: University of Oulu, 27-42.
Fairclough, N. (ed). 1992. Critical Language Awareness. London: Longman.
Kasper, G. 1995. "Wessen Pragmatik? Für eine Neubestimmung fremdsprachlicher
Handlungskompetenz". Zeitschrift für Fremdsprachenforschung 6: 69-94.
Kurzon, D. 1996. "The maxim of quantity, hyponymy and Princess Diana".
Pragmatics 6: 217-227.
Morris, C. 1964. Signification and Significance. Cambridge, MA: MIT Press.
Pennycock, A. 1994. "Incommensurable discourses". Applied Linguistics 15: 115-138.
Peräkylä, A. 1993. "Invoking a hostile world: Discussing the patient's future in AIDS
counselling". Text 13: 291-316.
Richards, I.A. 1930. Seven Types of Ambiguity. London: Chatto & Windus (Revised
edition 1961, Harmondsworth: Penguin).
Roberts, R.M. and Kreuz, R.J. 1993. "Nonstandard discourse and its coherence".
Discourse Processes 16: 451-464.
Rose, K.R. 1994. "On the validity of discourse completion tests in non-western
contexts". Applied Linguistics 15: 1-14.
Sanders, T., Spooren, W. and Noordman, L. 1992. "Toward a taxonomy of coherence
relations". Discourse Processes 15: 1-35.
Santos, T. 1992. "Ideology in composition: L1 and ESL". Journal of Second
Language Writing 1: 1-15.
Schiffrin, D. 1994. Approaches to Discourse. Oxford: Blackwell.
Stolze, R. 1992. Hermeneutisches Übersetzen. Tübingen: Narr.
Tannen, D. 1990. "Discourse analysis: the excitement of diversity". Text 10: 109-111.
Voss, J.F., Wiley, J., Ciarrochi, J., Foltz, P. and Silfies, L. 1996. "Race and the
representation of discourse: fictitious scenarios and the O.J. Simpson case".
Discourse Processes 22: 103-144.
Widdowson, H. 1995. "Discourse analysis: a critical view". Language and Literature
4: 157-172.
Widdowson, H. 1996. "Reply to Fairclough: Discourse and interpretation - conjectures
and refutations". Language and Literature 5: 57-69.
Wierlacher, A. 1992. "Toleranzforschung. Zur Forschungsplanung interkultureller
Germanistik". Jahrbuch Deutsch als Fremdsprache 19: 13-29.
Coherence and Meaning 265
Wolfson, N., Marmor, T. and Jones, S. 1989. "Problems in the comparison of speech
acts across cultures". In S. Blum-Kulka, J. House and G. Kasper (eds), Cross-
Cultural Pragmatics: Requests and Apologies. Norwood: Ablex, 174-196.
Zimmermann, P. (ed) 1989. Interkulturelle Germanistik: Dialog der Kulturen auf
Deutsch? Frankfurt: Lang.
A Bibliography of Coherence and Cohesion
Uta Lenk, Sarah Gietl and Wolfram Bublitz
Universität Augsburg
Anderson, R.C., Reynolds, R.E., Schallert, D.L. and Goetz, E.T. 1977.
"Frameworks for comprehending discourse". American Educational
Research Journal 14: 367-381.
Andersson, E. 1978. "An analysis of textual cohesion in a passage from Maria
Gripe's 'Hugo och Josefin'". In J.-O. Östman (ed), Cohesion and
Semantics. Åbo: Akademi, 131-143.
Andriessen, J., de Smedt, K. and Zock, M. 1996. "Discourse planning:
Empirical research and computer models". In T. Dijkstra and K. de Smedt
(eds), Computational Psycholinguistics. London etc. : Taylor and Francis,
247-278.
Aronsson, K. and Rundström, B. 1989. "Cats, dogs, and sweets in the clinical
negotiation of reality: on politeness and coherence in pediatric discourse".
Language in Society 18: 483-504.
Asher, N. and Lascarides, A. 1994. "Intentions and information in discourse".
Proceedings of the 32nd Annual Meeting of the Association of
Computational Linguistics June, 27-30, 1994. New Mexico.
Auer, P. 1986. "Kontextualisierung". Studium Linguistik 19: 22-47.
Auer, P. 1995. "Context and contextualization". In J. Verschueren, J.-O.
Östman and J. Blommaert (eds), Handbook of Pragmatics, 1995
Installment. Amsterdam: Benjamins. (19pp.)
Auer, P., Couper-Kuhlen, E. and Mueller, F. 1999. Language in Time: The
Rhythm and Tempo of Spoken Interaction. New York: Oxford University
Press.
Bäcklund, I. 1988. "To begin with, this is the problem, for example. On some
reader-oriented structural markers in English expository text". Studia
Linguistica 42: 60-68.
Bäcklund, I. 1990. "To sum up. Initial infinitives as cues to the reader". In G.
Caie et al. (eds), Proceedings from the Fourth Nordic Conference for
English Studies. Helsingør, May 11-13, 1989. Copenhagen: University of
Copenhagen, 289-302.
Ball, W.J. 1986. Dictionary of Link Words in English Discourse. London:
Macmillan.
Baltaxe, C.A. and D'Angiola, N. 1992. "Cohesion in the disclosure interaction
of autistic, specifically language-impaired, and normal children". Journal
of A utism and Developmental Disorders 22: 1-21.
Bamberg, M. and Marchman, V. 1990. "What holds a narrative together? The
linguistic encoding of episode boundaries". Papers in Pragmatics 4: 58-
121.
Banfield, A. 1978. "The formal coherence of represented speech and thought".
A Journal for Descriptive Poetics and Theory of Literature 3: 289-314.
270 Uta Lenk, Sarah Gietl, and Wolfram Bublitz
Black, J.B. and Bern, H. 1981. "Casual coherence and memory for events in
narratives". Journal of Verbal Learning and Verbal Behavior 20: 267-
275.
Blakemore, D. 1988. "The organization of discourse". In F. Newmeyer (ed),
Linguistics: the Cambridge Survey, Vol IV, Cambridge: Cambridge
University Press, 229-250.
Blakemore, D. 1992. Understanding Utterances. Oxford: Blackwell.
Bosch, P. 1989. "Coherence and cohesion: comments on Roger G. van de
Velde's paper 'Man, verbal text, inferencing, and coherence'". In W.
Heydrich, F. Neubauer, J. Petöfi and E. Sözer (eds), Connexity and
Coherence: Analysis of Text and Discourse. Berlin: de Gruyter, 218-227.
Braecke, C , Geluykens, R. and Pelsmaekers, K. 1997. "Clause ordering as a
text-building device in written L2". In M. Pütz. (ed), The Cultural
Context in Foreign Language Teaching. Amsterdam: Benjamins, 35-52.
Broek, P. van den. 1990. "Causal inferences and the comprehension of narrative
texts". In A.C. Graesser and G.H. Bower (eds), Inferences and Text
Comprehension. San Diego: Academic Press, 175-196.
Brown, E.K. 1980. "Grammatical incoherence". In H. Dechert and M. Raupach
(eds), Temporal Variables in Speech. The Hague: Mouton de Gruyter,
27-37.
Brown, G. and Yule, G 1983. Discourse Analysis. Cambridge: Cambridge
University Press.
Bublitz, W. 1988. Supportive Fellow-Speakers and Cooperative Conversations.
Amsterdam: Benjamins.
Bublitz, W. 1989. "Topical coherence in spoken discourse". Studia Anglica
Posnaniensia 22: 31-51.
Bublitz, W. 1994. "In the eye of the beholder: 'the rather mystical notion of
coherence'". In K. Carlon, K. Davidse and B. Rudzka-Ostyn (eds),
Perspectives on English. Studies in Honour of Professor Emma Vorlat.
Leuven, Paris: Peeters, 213-230.
Bublitz, W. 1996a. "'I bought a freezer - you've bought a freezer - they've
bought a freezer': repetition as a text-building device". In C. Bazzanella
(ed), Repetition in Dialogue. Tübingen: Niemeyer, 16-28.
Bublitz, W. 1996b. "Semantic prosody and cohesive company: somewhat
predictable". Leuvense Bijdragen 85: 1-32.
Bublitz, W. (forthcoming). "Cohesion and coherence". In J. Verschueren, J.-O.
Östman, J. Blommaert and C. Bulcaen (eds), Handbook of Pragmatics.
Amsterdam: Benjamins.
Bublitz, W. and Lenk, U. This volume. "Disturbed coherence: 'Fill me in'",
272 Uta Lenk, Sarah Gietl, and Wolfram Bublitz
Granger, S. and Tyson, S. 1996. "Connector usage in the English essay writing
of native and non-native EFL speakers of English". World Englishes 15:
17-27.
Gray, S. 1995. "Linear coherence and relevance: logic in computer-human
'conversations'". Journal of Pragmatics 23: 627-648.
Grimes, J.E. 1975. The Thread of Discourse. The Hague: Mouton de Gruyter.
Grinder, J. and Postal, P.M. 1971. "Missing antecedents". Linguistic Inquiry 2:
269-312.
Grosu, A. 1987. "On acceptable violations of parallelism constraints". In R.
Dirven and V. Fried (eds), Functionalism in Linguistics. Amsterdam:
Benjamins, 425-457.
Grosz, B.J. and Sidner, C.L. 1986. "Attention, intentions and the structure of
discourse". Computational Linguistics 12: 175-204.
Grosz, B.J. and Sidner, C.L. 1990. "Plans for discourse". In P.R. Cohen, J.
Morgan and M.E. Pollack (eds), Intentions in Communication.
Cambridge, MA: MIT Press, 417-444.
Grosz, B.J., Joshi, A.K. and Weinstein, S. 1995. "Centering: a framework for
modeling the local coherence of discourse". Computational Linguistics
21:203-225.
Gülich, E. and Kotschi, Th. 1983. "Les marqueurs de la reformulation
paraphrastique". Cahiers de Linguistique Française 5: 305-351.
Gumperz, J.J. 1982. Discourse Strategies. Cambridge: Cambridge University
Press.
Gumperz, J.J., Kaltman, H. and O'Connor, M.C. 1984. "Cohesion in spoken
and written discourse: ethnic style and the transition to literacy". In D.
Tannen (ed), Coherence in Spoken and Written Discourse. Norwood:
Ablex, 3-20.
Gutwinski, W. 1974. Cohesion in Literary Texts: a Study of some Grammatical
and Lexical Features of English Discourse. The Hague: Mouton de
Gruyter.
Haberlandt, K. and Bingham, C. 1978. "Verbs contribute to the coherence of
brief narratives: reading related and unrelated sentence triples". Journal of
Verbal Learning and Verbal Behavior 17: 419-426.
Hahn, U. 1985. "On lexically distributed text parsing. A computational model
for the analysis of textuality on the level of text cohesion and text
coherence". In F. Kiefer (ed), Linking in Text. Konstanz: Universität
Konstanz.
Halliday, M. A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Bibliography 281
Halliday, M.A.K. and Hasan, R. 1985. Language, Context, and Text: Aspects of
Language in a Social-Semiotic Perspective. Geelong: Deakin and 1989,
Oxford: Oxford University Press.
Hardy, D.E. 1996. "Topic versus cohesion in the prediction of causal ordering
in English conversation". Discourse Processes 21: 237-254.
Hartveldt, R. 1987. Pragmatic aspects of coherence in discourse. Groningen.
Harweg, R. 1968. Pronomina und Textkonstitution. München: Fink.
Harweg, R. 1989. "Is coherence a pragmatic phenomenon? Comments on
Dieter Viehweger's paper 'Coherence - interaction of modules'". In W.
Heydrich, F. Neubauer, J. Petöfi and E. Sözer (eds), Connexity and
Coherence: Analysis of Text and Discourse. Berlin: de Gruyter, 283-290.
Hasan, R. 1968. Grammatical Cohesion in Spoken and Written English.
London: Longman.
Hasan, R. 1984. "Coherence and cohesive harmony". In J. Flood (ed),
Understanding Reading Comprehension: Cognition, Language, and the
Structure of Prose. Newark: International Reading Association, 181-219.
Hatakeyama, K., Petöfi, J. and Sözer, E. 1985. "Text, connexity, cohesion,
coherence". In E. Sözer (ed), Text Connexity, Text Coherence. Aspects,
Methods, Results. Hamburg: Buske, 36-105.
Hawes, L.C. 1983. "Conversational coherence". In R. Craig and K. Tracy (eds),
Conversational Coherence: Form, Structure, and Strategy. Beverly Hills:
Sage, 285-298.
Heinemann, W. and Viehweger, D. 1991. Textlinguistik. Tübingen: Niemeyer.
Hellman, C. 1995. "The notion of coherence in discourse". In G. Rickheit and
C. Habel (eds), Focus and Coherence in Discourse Processing. Berlin: de
Gruyter, 190-202.
Heritage, J. 1984. "A change-of-state token and aspects of its sequential
placement". In M.J. Atkinson and J. Heritage (eds), Structures of Social
Action. Studies in Conversation Analysis. Cambridge: Cambridge
University Press, 299-345.
Heydrich, W. and Petöfi, J. (eds). 1986. Aspekte der Konnexität und Kohärenz
von Texten. Hamburg: Buske.
Heydrich, W., Neubauer, F., Petöfi, J. and Sözer, E. (eds). 1989. Connexity and
Coherence: Analysis of Text and Discourse. Berlin: de Gruyter.
Hickmann, M. 1980. "Creating referents in discourse: a development analysis of
linguistic cohesion". Papers from the Sixteenth Regional Meeting of the
Chicago Linguistic Society.
Hickmann, M. 1991. "The development of discourse cohesion: some functional
and cross-linguistic issues", In G. Piéraut-Le Bonniec and M. Dolitsky
282 Uta Lenk, Sarah Gietl, and Wolfram Bublitz
Petöfi, J. and Sözer, E. (eds). 1983. Micro and Macro Connexity of Texts.
Hamburg: Buske.
Pilus, Z. 1996. "Coherence and students' errors. Weaving the threads of
discourse". English Teaching Forum 34: 44-49.
Pit, M. 1997. "Subjective and objective construal in coherence relations". In J.
Don and T. Sanders (eds), Utrecht Institute for Linguistics. Yearbook
1997. Utrecht: Utrecht Institute for Linguistics, 45-70.
Polanyi, L. 1976. "The role of redundancy in cohesion and evaluative
functioning in narrative". Rackham Literary Studies 7: 19-38.
Polanyi, L. 1985. The Structure of Discourse. Norwood: Ablex.
Ragan, S. 1983. "Alignment and conversational coherence". In R. Craig and K.
Tracy (eds), Conversational Coherence: Form, Structure, and Strategy.
Beverly Hills: Sage, 157-171.
Rambow, O. (ed). 1993. Intentionality and Structure in Discourse Relations.
Proceedings of the ACL-Workshop, Ohio State University, Columbus,
Ohio.
Randquist, M. 1985. "The barely visible glue - some aspects of textual
connectedness". In E. Sözer (ed), Text Connexity, Text Coherence.
Aspects, Methods, Results. Hamburg: Buske, 189-218.
Redeker, G. 1990. "Ideational and pragmatic markers of discourse structure".
Journal of Pragmatics 14: 367-381.
Redeker, G. 1991. "Linguistic markers of discourse structure". Linguistics 29:
1139-1172.
Redeker, G. (forthcoming). "Coherence and structure in text and discourse". In
W. Black and H. Bunt (eds), Computational Pragmatics: Abduction,
Belief and Context. London: University College Press.
Reichman, R. 1978. "Conversational coherency". Cognitive Science 3: 283-327.
Reinhart, T. 1980. "Conditions for text coherence". Poetics Today 1: 161-180.
Rentel, V.M., King, M.L., Pappas, C.C. and Pettegrew, B.S. 1983. A
Longitudinal Study of Coherence in Children's Written Narratives. Final
Report to National Institute of Education. Columbus: The Ohio State
University Research Foundation.
Ribeiro, B.T. 1994. Coherence in Psychotic Discourse. Oxford: Oxford
University Press.
Rickheit, G. and Strohner, H. 1991. "Towards a cognitive theory of linguistic
coherence". Theoretical Linguistics 18: 209-237.
Rickheit, G , Günther, U. and Sichelschmidt, L. 1992. "Coherence and
coordination in written text: reading time studies". In D. Stein (ed),
Cooperating with Written Texts. The Pragmatics and Comprehension of
Written Texts. Berlin: de Gruyter, 103-127.
290 Uta Lenk, Sarah Gietl, and Wolfram Bublitz
Kaplan, R. 56; 88
keying 5; 126-145
Galtung, J. 88f
gambits 165 -L-
generic structures 111; 125
generic unity 111
Lakoff, G. 83
genre 37; 77-81; 85ff; 91, 96; 98; 100;
laughter 125-145
lOlf; 111f;119-122; 126; 129; laughter particles 17; 27; 32; 126; 133;
135; 145; 155; 165
137; 139; 141; 143
learner language 61; 72
Index 299
learners' strategies 4; 57; 64ff; 146 prosodic: ~ animation 12; 23; ~ cues 4;
lexical field 66 11-30; 125; 127; 175;~ effects 14;
lexico-grammatical deficits 56 ~ signals 18; ~ variation 3
linguistic maturation/ maturity 57; 59f punchline 125; 127f; 139; 133; 136;
logical connectors 56f; 200 139; 145
-M- -Q-
metapragmatic 27; ~ function 126; 147 questions 4; 37-54; 123; 163f; 194-200
metatext, means and functions 199;
201 -R-
mind 216; 252f; 263
misunderstanding 97; 175ff reference 5; 116; 121;125; 154; 165ff;
modulation 120 172; 189
multi-level thesis 236-245 register 6 57; 62; 67; 69; 71f; 106;
llff; 119; 153; 173
-N- register break 5; 154; 172
relations: classification of ~ 240f;
Natural Language Generation 236 logical ~ 55; 57; propositional ~
negotiation of topic, see negotiation of 242; 247; rhetorical ~ 236f
coherence relevance 11; 47; 118f; 127; 129; 236;
nominal style 64 242
non-coherence, see incoherence repair 6; 11-30; 176; 181; 183ff
non-native discourse 4; 55-73; 101; reported speech 4; 11-30
103 Rhetorical Structure Theory 236
non-nativeness, see non-native rhythm 19ff ; 167
discourse routines 254
-O- -S-
-T-
-U-
-V-
-W-
-Y-
Yeltsin, B. 77