I know that women don’t like me!’ Presuppositions in therapeutic discourse Marta Andersson pot

‘I know that women don’t like me!’ Presuppositions in
therapeutic discourse
Marta Andersson
English Language Department, Stockholm University, Universtietsva
¨
gen 10E, 10691 Stockholm, Sweden
Received 23 March 2006; received in revised form 11 March 2008; accepted 25 September 2008
Abstract
One of the biggest problems concerning presuppositions has been correctly dealing with their sensitivity to the context, i.e. why
inferences triggered by certain expressions do not project out in all linguistic environments, even though the triggering words
preserve their semantic content in different settings. The answer which is of particular interest here goes along with the principles of
the binding theory of presuppositions developed by van der Sandt (1992). According to this theory, presuppositions behave as
anaphors and can be resolved in the same way at the level of discourse representation. This article contributes to a very scarce body
of empirical work on presuppositions, as it scrutinizes examples of presuppositions that act like discourse anaphors in the context of
three psychotherapeutic sessions. Such sessions can be analyzed in the same way as ordinary spoken discourse; however, the initial
premise that the usage of presuppositions differs in this genre in comparison to daily interaction is confirmed. The results of both
quantitative and qualitative analysis indicate that presuppositions are used for different strategic reasons in the two genres
compared, which influences the way they should be interpreted and also their frequency.
# 2008 Elsevier B.V. All rights reserved.
Keywords: Anaphors; Accommodation; Binding; Presupposition; Representation
1. Introduction
The crux of van der Sandt’s (1992) binding theory, proposed as a solution to the projection problem of presuppositions
(discussed further), is the idea that they can be handled using the same mechanism that is used to resolve anaphoric
pronouns in Discourse Representation Theory (DRT, Kamp, 1981; Kamp and Reyle, 1993). Van der Sandt’s claim is that
presupposed information can be bound to an already existing antecedent in the discourse, exactly like pronouns.
Therefore it can serve as one of the ways of referring to such an antecedent, along with pronouns and demonstratives.
However, as van der Sandt argues, even if no suitable antecedent is given in the discourse, the rich semantic content of
presuppositions enables them to accommodate, i.e. create their own antecedent, which is impossible for anaphors or
pronouns.
The biggest dissimilarity between these two strategies for presupposition resolution is that bound presuppositions (and
anaphors) refer to hearer-known information already present in the discourse, while accommodation is regarded as a way

to introduce discourse-new information, in which case the hearer needs to adjust cognitive structures and create a part of
the speaker’s meaning by using inferences. This is feasible on the basis of presumed common knowledge (Stalnaker,
1974:450) that enables the speaker to surmise what the hearer possibly knows and introduce such information felicitously.
www.elsevier.com/locate/pragma
A
vailable online at www.sciencedirect.com
Journal of Pragmatics 41 (2009) 721–737
E-mail address:
0378-2166/$ – see front matter # 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.pragma.2008.09.033
Most of the work on presuppositions written before van der Sandt’s (1992) binding theory was coined had examined
them merely in meticulously contrived examples eagerly quoted and dissected from paper to paper (see Karttunen,
1973; Gazdar, 1979; Levinson, 1983). The only empirical studies on presuppositions known to me are those by Prince
(1978) on it-clefts, Fraurud (1990) on definite descriptions, Sbisa
`
(1999) on the persuasive and informative use of
presuppositions in the Italian press and Spenader (2002) on presuppositions in casual and casual-like conversation.
Apart from that the approach to presuppositions has been mainly philosophical rather than discourse-analytic,
although, as Spenader (2002:49) argues, looking at naturally produced examples can yield new results in research into
the communicative effects of using these inferences. As features related to the context, presuppositions disclose the
speaker’s state of knowledge and his/her assumptions on the hearer’s knowledge. Thus they cause various reactions in
the hearer, which is particularly discernible when analyzing longer stretches of spoken dialogues. Hence, from the
vantage point of the speaker’s intentions, therapeutic discourse seems to be a particularly interesting subject for
analysis, as the exchange of information in this genre has a clear pragmatic goal, unlike in casual conversation (often
just ‘talking for the sake of talking’, Eggins and Slade, 1997:6) investigated by Spenader. Moreover, the information
communicated via presuppositions and the way in which they contribute to the hearer’s understanding of the message
are particularly salient in therapeutic discourse, as they can influence and eventually even change people’s (patients’)
behaviour, which is not always the case in casual talk. We can, therefore, presume that the strategies for presupposition
resolution may be different in these two genres.
In this article I will investigate 1) whether binding as a strategy for presupposition resolution is more frequent in

therapeutic discourse than in casual conversation. Such an assumption seems rather apt, as it relates to constant
referring to the same limited and repeatedly examined issues (ubiquitous in therapeutic sessions), and therefore
suggests that one of the features distinguishing therapeutic discourse from ordinary conversation might be the
frequency of bound presuppositions. Moreover, as Labov and Fanshel (1977:62) write, the aforementioned acts of
reference play a salient role in making the flow of a conversation smooth, which seems particularly desirable in
therapeutic discourse and will also be illustrated in the further discussion.
Van der Sandt’s binding theory also brings in part into question what the utility of using a bound presupposition
would be, if an anaphor would suffice. In section 2.2 I will indicate 2) that in many cases the rich semantic context of
presuppositions validates the aptness of their use as more efficient counterparts of anaphors and also 3) that the use of
presuppositions instead of anaphors creates certain rhetorical effects particularly useful in therapeutic discourse.
The following paper will 4) also attempt to investigate instances when there are no accessible reference markers in
the preceding discourse. In such cases presuppositions are usually accommodated, but the kind of information they
introduce may create certain obstructions in communication, particularly when the information is hearer-new. Hence
my hypothesis is that introducing new information via presuppositions may be avoided in the genres that have a clearly
specified pragmatic goal, which will be discussed in section 2.3.
1.1. Presuppositions in the binding theory
From the vantage point of their logical properties presuppositions are inferences induced by certain linguistic
expressions (presuppositional triggers). They can be distinguished from the context they occur in by means of
established linguistic tests (Levinson, 1983:168). Consider the following examples:
(1) Eve
has stopped smoking.
(2) Eve
hasn’t stopped smoking.
We can infer both from (1) and (2) that Eve has smoked before, as the underlined expressions implicate this fact. The
negation test illustrated in (2) is a common way in which inferences can be regarded as presuppositional. This is converse
to asserted information and entailment, for the example in (1) entails the fact that Eve does not smoke and asserts that she
quit, all of which disappear under the negation. However, certain contexts make presuppositions vanish:
(3) If Eve has smoked, she
has stopped smoking.
The presupposition that was triggered in (1) and (2) obviously does not project out of this environment, as the fact of

Eve’s smoking is questioned by the antecedent of the conditional. This illustrates the most perplexing aspect of the
M. Andersson / Journal of Pragmatics 41 (2009) 721–737722
nature of presuppositions, which, in the literature, is called the projection problem (see Karttunen, 1974; Levinson,
1983). This problem concerns presuppositions triggered in compound sentences in which the implications of separate
clauses do not always emerge (as in (3), which indicates their sensitivity to the context of an utterance
(Geurts, 1999a:46).
The projection problem had for decades been treated very differently by linguists who adhered to semantic
approaches to presuppositions and those who subscribed to a pragmatic view (see Karttunen, 1973, 1974; Stalnaker,
1974; Gazdar, 1979; Levinson, 1983). A new light on this issue was shed by Stalnaker (1974), who claimed that
presuppositions treated from the pragmatic view of the speaker can differ from those treated semantically. According
to his intuitions pragmatic presuppositions do not hold in compound sentences, as the speaker builds on what s/he has
already said. Hence, what is asserted cannot be later presupposed, as in the case of conditional sentences where
presuppositions do not project out if information is asserted in the antecedent. Stalnaker was, therefore, one of the first
who treated presuppositions from the vantage point of context.
It is now generally acknowledged that the interpretation of presupposed information should not be tackled on the
level of the sentence, but we should instead account for the whole array of cognitive activities that accompany the
process of comprehension and interpretation, within which meaning and its representations are constructed by the
entire discourse. This goes along with the assumptions of DRT, which is a cognitivist theory according to which mental
processes involved in language comprehension play a prominent role. The theory is related to the representations of
the entire discourse (Discourse Representation Structures—DRS), not just individual sentences, and considers the
construction of meaning as a process that proceeds sentence by sentence. Each next sentence S contributes new
information that updates the already existing representation (DRS) and changes it into a new one that includes
S(Kamp and Reyle, 1993:11).
As a solution to the projection problem of presuppositions van der Sandt (1992) extendedDRTandproposedthe
binding theory, according to which presuppositions can be treated like pronouns or other anaphoric expressions at the
level of representation in DRT. Presuppositions can be resolved against the content of an evolving discourse by
establishing a link between them and their antecedents, and this one of the two possible mechanisms for
presupposition resolution is called binding. The other mechanism is based on the rich semantic content of
presuppositions thanks to which they can accommodate an antecedent even if it is not given in the discourse. This
constitutes, according to van der Sandt, the only essential difference between presuppositions and anaphors. Consider

first an example of binding:
(4) If Eve is seeing a guy, then
he is divorced.
(5) If Eve is seeing a guy, then
the guy is divorced.
As we can observe here, whether we use an anaphoric expression (the pronoun in (4)) or a presuppositional one
(i.e. the noun
1
in (15)), there will be no differences in interpretation of these sentences, and this, as van der Sandt
(1992:344) writes, suggests that there is a similar force that underlies both pronoun resolution and presupposition
projection. Hence we should not treat the presupposition that does not project in (5) as cancelled, but bound to an
antecedent in the preceding discourse, exactly like the pronoun in (4). In such a case the presuppositional construction
will be absorbed in its antecedent and lose its presuppositional status.
2
The anaphor in (4) is the most basic and the most uncontroversial example of incremental interpretation and it is, at
the same time, closely related to familiar patterns of cohesion (Halliday and Hasan, 1976:276). As we can observe in
(5), the presuppositional expression acts exactly in the same way, which makes the anaphoric character of definite NPs
and the presuppositions they mark uncontroversial.
It is not only definite NPs, however, that can be resolved by the same mechanism that DRT uses with anaphors. Van
der Sandt’s (1992:342) claim also covers aspectual and factive verbs, cleft sentences, temporal clauses and adverbs
like too or even, all of which can also be bound to an antecedent. Presuppositional information, according to van der
Sandt (1992), should be bound to an antecedent in discourse in order to get an interpretation, exactly like anaphors.
However, there is a difference between the two that emerges when a compatible antecedent cannot be identified in the
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 723
1
Definite NPs trigger a presupposition that there is such a referent given in the discourse.
2
This is why bound presuppositions should rather be called: presuppositional inferences, as van der Sandt argues. However, these two terms are
used interchangeably in the literature on presuppositions, which I will follow in the present study.
preceding discourse. In such a case a sentence with a pronoun will be impossible to interpret,

3
while a presuppositional
expression, as it has a richer descriptive content, will provide its own antecedent, add itself to the main context and
bind with that antecedent. This process is called accommodation. A good example of a presupposition that has to be
accommodated is given in Lambrecht (1994:14) in his discussion on markedness in information structure. The
example depicts a woman who causes a delay of a bus and apologises the other passengers saying
4
:
(6)
My car broke down.
The definite possessive NP (my car) presupposes that the speaker has a car. In van der Sandt’s (1992) view this
expression has sufficient semantic content to establish its own antecedent and bind with it. As he also argues, binding
just updates the information already present in the discourse, while accommodation revises the context of an utterance
and then enables an interpretation of presupposed information. As Spenader (2002:29) writes, the hearer has to infer
the context that the speaker is tacitly asking him/her to treat as already discourse-given, create part of the speaker’s
meaning and modify his/her own discourse representation by adding the presupposed information to the DRS. This
process is regarded as the speaker’s ‘mistake’ that the hearer needs to repair. Therefore accommodation is considered
to be a repair strategy and should be used only after all potential possibilities for binding fail (van der Sandt, 1992;
Geurts, 1999a).
1.2. Therapeutic discourse
A therapeutic interview can be described as a type of conversation (Labov and Fanshel, 1977:30). However,
a therapeutic session, unlike a casual conversation, is constrained by the agenda and the limitations of possible topics
to discuss. Therapeutic talk is a complex interactive situation between the patient whose task is to give the therapist an
account of his/her problem and the therapist whose job is to collect a valid database, develop an evolving
understanding of the patient and influence his/her way of thinking (see Blanchet et al., 2005:125ff). The process of
influencing derives from concrete verbal and non-verbal actions and interactional approach which should lead the
patient to reshaping his/her representation in a more efficient and logical manner. This means that the therapeutic
interventions should help the patient develop a new conception of his/her problem, i.e. insight (Blanchet et al.,
2005:126). The expected new representation may come only through further talk and gradual modifications to the
patient’s discourse and, eventually, his/her self-concepts.

The most problematic feature of therapy emerges at this point, as in the course of an ordinary talk we can
expect to be experts at least in one area: our inner world (Labov and Fanshel, 1977:34). In a therapeutic session the
patient’s biography often renders a substance that is moulded in a way that is divergent from what his/her own
model of this world represents. In order to influence this process the patient formulates his/her thoughts so that
they fulfil a certain emotive function: affect the listener, make him/her respond, move him/her, etc. These verbal
strategies pertain to pragmatic intricacies of language use that concern the speaker’s intended meaning in a certain
context. In the context of a therapeutic talk this meaning is a means to represent the patient’s model of
experiences.
In this model the patient makes certain assumptions (Bandler and Grinder, 1975:52). One of the ways in which they
protrude linguistically is presuppositions. These are a prominent characteristic of human discourse in general, but the
strategies for their resolution may be different in different genres. In therapeutic discourse presuppositions may serve
(from both the patient and the therapist’s perspective) as technical tools whose objective is to maintain the overall
continuity of the topic and enhance the discourse flow, but also as rhetorical devices used in order to influence the
interlocutor, provide him/her with additional clues, enhance the claim, or, possibly, serve as avoidance or defence
strategy. As the above reasoning indicates, therapeutic discourse is a clearly goal-oriented genre and therefore it is
interesting to investigate whether its presuppositions have different pragmatic functions from those generated in the
course of casual conversation.
M. Andersson / Journal of Pragmatics 41 (2009) 721–737724
3
This is, certainly, an oversimplified approach, since pronouns provide some descriptive content and in certain environments it is possible to
interpret a pronoun without any clear antecedent in the preceding context. Yet this discussion is beyond the scope of the present paper.
4
Neither the question of markedness of information, nor any other communicative and pragmatic functions of this example are the subject of the
discussion in this study.
1.3. Corpus and method
The corpus of this study consisted of transcripts of three authentic therapeutic sessions. Sessions 1 and 2 were
borrowed from therapeutic books (Session 1 from: Sluzki, 1990; and Session 2 from: Bandler and Grinder, 1975). Both
of them are initial meetings between the therapists and the patients, in Session 1 the patients are a couple (Jennifer and
Larry), while Session 2 has a single patient (Ralph). Session 3 was a subsequent meeting which I was lucky enough to
obtain a recording of thanks to the courtesy of an English-native-tongue therapist and her patient from The Institute of

Psychodrama, Sociometry and Group Psychotherapy in Stockholm.
5
The transcriptions were stored electronically, and all triggers were identified and analyzed in the context from the
vantage point of presupposition resolution, i.e. either a suitable antecedent was found in the possibly closest context or
presupposition was treated as accommodated. The triggers considered here were borrowed from Geurts (1999a:2) and
Spenader (2002:58) and comprise:
 definite NPs (the N; demonstrative NPs, possessives)
 it clefts (it is/was Ci that/whom/which S-Ci)
 aspectual verbs (begin, finish, start, stop, cease, continue, carry on, avoid, force, hesitate, prevent)
 factive verbs (know, realize, resent, find out, discover, notice, see)
Both quantitative and qualitative characteristics of the presuppositions triggered by specified expressions in the
context of this particular genre were, subsequently, compared with a corpus study of presuppositions in English
conversation by Spenader (2002).
Features excluded from the analysis comprise certain instances of definite NPs such as fixed expressions (e.g. in my
eyes), idioms (e.g. the nuts and bolts etc.), calculable forms (i.e. using information about a reference time and a
reference place, such as: in that moment), known information (entity that is new in the text, but belongs to general
knowledge, e.g. the world ) and also cases of bridging.
6
Examples of factive verbs devoid of that sentential complement
or at the end of a tone unit were also discarded (e.g. ‘‘She knew ’’ where it was hard to determine what ‘knew’ actually
presupposes). All this follows the procedure of Spenader’s (2002) study. However, abstract triggers in sentences of all
types were considered here, while Spenader’s survey was limited to those occurring in declarative sentences only.
The reason for that is that certain interrogative forms may make it rather difficult to get clear intuitions about what the
presuppositions in a sentence are; for instance: ‘Did you discover that John was the murderer?’ might presuppose that
John was the murderer, but could be a question about the presupposition itself. Such ambiguous examples were also
discarded in the present paper, but more straightforward interrogative forms were included, as they are crucial in
therapeutic discourse (see examples from Session 2 in 2.2 and 2.3). Presuppositions triggered in such non-declarative
sentences considered here constitute 3% of the whole corpus.
2. Analysis
2.1. Cases of binding and accommodation in numbers

Table 1 shows the strategies for resolution of definite NPs.
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 725
Table 1
Resolution of definite NPs in therapeutic discourse and casual conversation.
Trigger type Therapeutic discourse (7 900 words) Casual conversation (8 972 words)
Binding Accommodation Binding Accommodation
The NPs 70 51 72.5 96
Demonstrative NPs 33 14 21.5 9.5
Possessive NPs 59 49 31 56.5
Total 162 114 125 135
5
Stockholms Institut fo
¨
r Psykodrama, Sociometri och Gruppsykoterapi.
6
Cases like: ‘‘Sarah is married. The guy is divorced’’ (Session 3, line 18) where the definiteness of the nominal in the second sentence implicates
that it is a part of given information, even though physically absent in the preceding utterance. According to the binding theory it can be treated as a
stage between binding and accommodation (see also Clark, 1975).
The figures for definite NPs in casual conversation are based on a subcorpus of three dialogues, which had a
combined length of 8972 words. These figures are the average of the results of the two annotators annotating
Spenader’s corpus of definite NPs
7
for binding and accommodation (for practical reasons I did my own annotation).
Therefore the corpora sizes are very comparable.
There is a statistically significant difference at p < .005 between the number of bound and accommodated NP
presuppositions in therapeutic talk. This observation is revealing, as it indicates that therapeutic discourse is, indeed,
different from casual conversation. Particularly the balance between the bound and accommodated possessive NPs in
therapy talks in comparison to daily discourse is significant, as they refer to constantly repeated personal information
(e.g. ‘my husband’) ubiquitous in this genre. Perhaps the most surprising result is the higher frequencies of both bound
and accommodated ‘the NPs’ in casual conversation (the difference is significant at p < .001). Yet therapeutic

discourse is fixed on the same, recurrent topics (it may, therefore, have more pronouns), while in ordinary interaction
shifts in topic are quite frequent (Spenader, 2002:51-ff). Thus the variety of both referring and new expressions can be
higher in the latter genre. Table 2 shows the results for abstract triggers.
Identifying a suitable resolution strategy was probably less complex for the analyst in the present study than in the case
of casual conversation, where the information is more messy and diffused among several speakers (Spenader, 2002:68).
However, it was not always easy to decide if information should be regarded as hearer-new or maybe discourse-new and,
in the case of binding, if certain information was semantically similar enough to be an antecedent of another expression.
Yet considering the sizes of both corpora we can conclude that abstract presuppositional triggers are substantially more
frequent in therapeutic discourse than in casual conversation (the difference is significant at p < .001). The reason might
be that they can be used to refer precisely to abstract objects (facts, opinions, situations), which is crucial in this genre.
Table 3 shows the number and distribution of all the chosen presuppositional triggers in the material analyzed.
The figures in Table 3 indicate that binding predominates in Session 1 ( p < .001) and 2 ( p < .005), while in
Session 3 both strategies for presupposition resolution are used quite freely (with a non-significant predominance of
accommodation). Since Session 3 is the only analyzed transcript of a subsequent therapeutic talk, while the other two
are first interviews, it can be suggested that there might be certain differences in the use of presuppositions between
initial and subsequent therapeutic sessions. A comparison between these two types of therapeutic talk, however,
M. Andersson / Journal of Pragmatics 41 (2009) 721–737726
Table 2
Resolution of abstract triggers in therapeutic discourse and casual conversation.
Trigger type Therapeutic discourse (7900 words) Casual conversation (233 000 words)
Binding Accommodation Binding Accommodation
It-clefts 9 4 19 12
Aspectual verbs 14 17 10 58
Factive verbs 11 6 28 81
Total 34 27 57 151
Table 3
Resolution of each trigger type in each session.
Binding Accommodation Binding Accommodation Binding Accommodation
The NPs 48 23 1 3 21 25
Demonstrative NPs 16 3 3 2 14 9

Possessive NPs 28 29 9 1 22 19
It-clefts 4 – – – 5 4
Aspectual verbs 2 1 4 1 8 15
Factive verbs – 1 6 – 5 5
Total 98 57 23 7 76 77
7
For more details see Spenader (2002:162-ff).
deserves a separate study based on more empirical material. Yet the frequencies presented here will be interpreted in
more detail in section 3.
As the above results show, binding is overall the predominant strategy for presupposition resolution in the analyzed
corpus. Yet more material is needed in order to provide even stronger empirical evidence for this initial observation for
the whole genre. This concerns particularly the strategies of resolution of abstract triggers in whose case a bigger
corpus is needed in order to obtain significant results.
2.2. Bound presuppositions in the context
This section will show that presuppositions act like their anaphoric counterparts and are interchangeable with
anaphors but also often show much greater adaptability in the context, i.e. they are able to pick up an antecedent even if
it would not be possible for a pronominal expression (e.g. due to remoteness). They also contribute to the overall
continuity of a topic, which is the reason why they are often used instead of anaphors. The continuity of the topic is a
feature of particular importance in therapeutic discourse, but the examples below will also illustrate rhetorical effects
and other possible reasons for using full-fledged presuppositions instead of anaphors in this type of conversation.
Moreover, in more complex cases the mechanism of binding will also be described in closer detail. Triggered
information will be
underlined and its antecedents italicised in the whole paper.
There is only one example in my corpus that illustrates a presupposition that, in terms of traditional presupposition
theory, does not project out the embedded context:
(7) definite (possessive) NP; induced information: specified feelings (S3:135–139)
8
P: ( ) He deprived us of all this and he still says that he had feelings for me, you know, loved me; cared
I don’t believe him. In my eyes even if he had feelings for me (a),
those feelings (b) didn’t have anything in

common with love (A)
The sentence (A) could serve as a paradigm example in the literature on presuppositions written a few decades ago.
As we can see, the presupposition:
those feelings is not inherited by the whole compound sentence, since the
demonstrative points at feelings whose existence is questioned in (a). According to the previous theoretical approach
the presupposition is cancelled here, while van der Sandt (1992) claims that it acts exactly like an anaphoric expression
(‘they’) would – it is bound in the first conjunct.
The whole utterance, rather perplexing for linguists and language philosophers, in this context simply shows that
the patient does not commit herself to the assumption that her husband loved her at the moment she refers to and the
full-fledged presupposition (instead of anaphor also in this context felicitous) supports her belief state. Consider an
example of presupposition that can be replaced with an anaphoric expression:
(8) factive; TP:
9
it is safe to have these feelings (S1:37–39)
J: Well, when I say ‘‘communication’’, I mean exchange of feelings and ideas and thoughts, you know, a sharing of
these things. I mean exposing one’s vulnerabilities and feeling that it’s safe to do so (a). And my feeling is that if
you can do that with somebody, you establish a closer relationship with that person.
T: That makes sense.
J: And, uh, that’s what I mean by communication. And, also, not being afraid to show these feelings is important.
I think that’s the first step in
realizing that it’s OK or it’s safe to have these feelings and to show them to the other
person (b).
T: Are you talking about yourself, or are you (Sluzki indicates ‘‘gotcha’’ by making a stabbing gesture toward
Larry and simultaneously emitting a short, humorous noise.)
J: Oh, I’m talking more about Larry.(c)
The triggering expression in (b) (
realize) and its complement are a reformulation of an earlier utterance. The patient
decided on using a presupposition, although the entire sequence in (b) could have been replaced with an anaphor
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 727
8

S3:135–139—Session 3, lines: 135–139.
9
Abbreviation for: triggered presupposition.
(‘this’). Yet this simple expression might have not been specific enough to clearly delimit what part of her concept is
referred to. Another reason for using a full-fledged presupposition might be the emphasis she wanted to put on her
views, which would not have the same rhetorical effect in the case of an anaphor. The presupposition makes her claim
in (a) clearer and stronger, which is particularly interesting when juxtaposed with the fact that the proposition is in the
scope of the propositional attitude verb: ‘think’ (Biber and Finnegan, 1989:119). This verb shows epistemic hesitation
by the speaker on the one hand, which, on the other hand, is immediately suppressed by the presence of the factive
verb. According to the mainstream conviction, when using a factive verb, the speaker presupposes that the complement
of the sentence expresses a true proposition (Kiparsky and Kiparsky, 1971:345; Levinson, 1983), i.e. commits him/
herself to the truth of this proposition, unlike in the case of epistemic stance features marking hesitation.
10
Since
Jennifer almost immediately admits that her utterance concerned her ideas about how her partner should act in their
relationship (c), we can conclude that the stance marker was used probably in order to hedge the force of
presupposition triggered by the factive. Consider another example:
(9) it-cleft; TP: there is a wall built (S3:50–54)
P: No, maybe it just sounds like that I love Jonas, if I didn’t, I wouldn’t be here, I wouldn’t even try It’s just that
after so many years of quarrels, I feel that there’s a wall (a) between us. But I didn’t want to build it (b), I don’t
feel guilty,
it’s him who built up this wall (c)
T: Do you ever try to do something about it? About the wall?
As Spenader (2002:81) argues, it-clefts are particularly well suited for corrections, as they focus on one element
(the one that must be corrected) and background the rest. In therapeutic discourse corrections seem to serve the
obvious purpose of assigning the responsibility to the other partner and defending oneself (Labov and Fanshel,
1977:229). Here the patient introduces a new referent in (a), develops the representation in (b)
11
and binds (c) to the
information provided both in (a) and (b). The anaphoric relationship between (a) (b) and (c) is discernible on the basis

of the correction that the latter utterance makes to the first two. The presupposition triggered by the it-cleft in (c) points
to something that might have been mistakenly inferred from both (a) and (b) and corrects it. The full presuppositional
expression, again, could have been replaced with a simple ‘did it’. In such a case, however, the whole utterance would
have lost its emphasis and rhetorical effect. Furthermore, the obvious and manipulative focus on the other person’s
guilt would not be so conspicuous and, at the same time, the patterns of problematic behaviours not so easy to perceive.
In this case the therapist immediately identifies the patient’s assumptions of being a victim of her partner and
challenges them by shifting the focus to what she does in the relationship. Thus the patient’s strategy actually
facilitates what the therapist wants to achieve (i.e. changes in the patient’s self-concepts). Consider another example:
(10) factive; TP: Janet was not interested in Ralph (S2:34–41)
R: She just didn’t like me.(A)
T: How, specifically, do you know that Janet didn’t like you?
R: She wasn’t interested in me.(a)
T: Interested in what way?
R: She didn’t pay attention to me.(B)
T: How didn’t she pay attention to you?
R: She didn’t look at me.
T: Let me see if I understand this. You
know that Janet wasn’t interested in you (b) because she didn’t
look at you?
R: That’s right! (c)
T: Is there any way you could imagine Janet not looking at you and still be interested in you?
R: Well I don’t know
T: Do you always look at everyone you’re interested in?
R: I guess not always. (d)
M. Andersson / Journal of Pragmatics 41 (2009) 721–737728
10
‘Think’ is a doubt verb (see Biber and Finnegan, 1989; Ajimer, 1996).
11
Note that anaphoric ‘it’ is used here only in the relation to (a).
The presupposition in this passage could easily have been replaced by ‘this’ (like in (8)), but this simple anaphoric

expression might have not been specific enough to clearly demarcate what part of the patient’s concept is referred to:
(A), (a) or (B). Even though all these utterances convey similar information, relevant reference seems to be particularly
important in therapeutic discourse, as the therapist shows.
The therapist’s intervention in the quoted passage consists in repeating verbatim the patient’s generalized and
scanty information and bolstering it by means of presupposition (
know that Janet wasn’t interested in you). This
eventually leads to the patient corroborating his views in (c) and to the reconstruction of the situation that is obscurely
introduced in the preceding discourse. The most important factor here, however, is the fact that the presupposition is
triggered by a factive verb (‘know’), which, as mentioned before (see (8)), presupposes that the uttered proposition is
true. Hence the patient’s loose ideas implying that he knows what other people think and feel about him, summarized
and bolstered by means of exactly this presupposition, provide a basis for the therapist’s attempt to make him see his
erroneous way of perceiving his personal problems (d). As Blanchet et al. (2005:124) write, the paradigm for all
psychotherapy is to encourage the patient to change his/her viewpoint and discourse about a subject of direct concern
to him/her. The above example illustrates how a deconstruction of the patient’s initial point of view may be initiated by
the therapist, in this specific case by means of a presupposition. Consider one more example:
(11) aspectual verb; TP: there is something to tell (S2:97–102)
T: Did you ever let her know for sure that you loved her [Ralph’s mother]? (a)
R: She knew (b)
T: How do you know she knew?
R:I I I guess I don’t.
T: What
prevents you from telling her? (c)
R: Ummm umm, maybe nothing.
T: MAYBE?
R: Well, I guess I could.
The aspectual verb
prevent triggers a presupposition that Ralph has something to tell to his mother. Ralph’s
utterance in (b): ‘she knew’ implies quite clearly that he did not tell his mother about his feelings and even though he
chooses not to spell it out, the therapist apparently obtains a clue that is strong enough to generate such a hypothesis.
The latter is expressed via an aspectual verb that presupposes that something stops Ralph from making the confession

in question. By saying it the therapist in a clear and concrete manner encourages the patient to modify his approach to
this sensitive issue, which he eventually does. The presupposition also suggests that the therapist is qualitatively
listening and actively exploring the patient’s conceptions of his problems.
Moreover, the presupposition is more felicitous in this context than anaphora would be (i.e. ‘what prevents you
from doing this?’), since it is the former expressions that make the utterance clearer, concrete and more informative, all
of which are crucial for establishing connections and understanding between therapists and their clients (see also
Hilsenroth and Cromer, 2007:210ff).
The above examples indicate that bound presuppositions are interchangeable with anaphors; however, using the
former can change the rhetorical effect and efficacy of utterances, while the latter lack the same force. Presuppositions
also exhibit greater adaptability in the context than anaphors do, as the next example will show:
(12) definite description; TP: shame (S3:21–155)
P: Shame, (a) shame is the punishment for that I’m ashamed because I tolerate what he did ( )
(58 turns later)
P: Yeah And when he almost managed to convince me that he was honest about his feelings, this huge punch
came Have you ever reflected on this word ‘‘dumped’’, how nasty it really is? You really start to feel like
trash when it concerns you!
T: Then you feel
the shame (b).
P: Exactly.
The most significant feature of the above passage is that the presupposition picks an antecedent at a considerable
distance and after the topic of the discussion has drifted away from what the patient initially said. This would not be
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 729
possible for an anaphoric expression, as its semantic content is not rich enough to bind over such a long distance, which
illustrates how presuppositions contribute to the discourse flow. In this example the presupposition in (b) gives the
therapist a chance to be perceived as a good and active listener interested and involved in the patient’s experiences.
Particularly Sessions 1 and 3 are pervaded with this type of long-distance bound presuppositions made by the
therapists, by which they can prove that they analyze and obtain clues when listening to the patients’ discourse.
Section 2.2 discussed the problems that van der Sandt’s (1992) binding theory has not yet satisfactorily addressed,
i.e. what the distinctive features are of cases where a full-fledged presupposition is used even though an anaphoric
expression would have functioned. As the above examples indicate, these uses are often occasions when an abstract

concept is being referred to, as in (8), (10), (11) or when a simple anaphoric expression is not specific enough, e.g. (8),
(10). Presuppositions can also be used instead of anaphors for rhetorical purposes, as in (8), (9), (10), (11), and, in some
settings, as more efficient counterparts of anaphoric expressions e.g. (12).
Section 2.3 will consider examples of presuppositions that do not have any antecedent to bind to.
2.3. Accommodation
Accommodation is considered to be a repair strategy, a ‘mistake’ that the speaker made when planning and making
his/her utterance. It is believed to be usable for signalling hearer-old and discourse new information, since this type of
information is also part of the common ground between interlocutors and simply makes the hearer recognize and
retrieve what is important in the current discourse (Spenader, 2002:111). Hearer-new information, on the other hand,
does not have this status and therefore is a potential source of communicational obstruction.
However, according to Spenader’s (2002:102-ff) findings about casual conversation, in surprisingly many cases
when the speaker believed that the information was hearer-new, s/he intentionally used presupposition, in order to
highlight the newly introduced entity. This observation is revealing, since the mainstream assumption concerning
presuppositions is that they are rhetorically related to discourse and serve a background function to other non-
presupposed information (see Asher and Lascarides, 1998; Delin, 1995). However, as Lambrecht’s (1994:61ff)
discussion on presuppositions triggered by factive verbs illustrates, in some cases the proposition expressed in the
complement clause does not have to be part of shared knowledge, for instance: I didn’t realize that you LIED to me!
This sentence presupposes that the speaker was lied to and asserts that s/he was not aware of this fact at some earlier
point in time, thus it could not have been part of common knowledge. Yet in this case such an interpretation is strongly
related to the accent placement, while Spenander’s study did not include any analysis of the influence of prosody on
presupposed material. This fact suggests that presuppositions may be in general (i.e. independently of prosodic
marking) quite commonly used to signal hearer-new information.
Since obstructions in communications are particularly infelicitous in goal-oriented genres, it is therefore assumed
that introducing new information by means of presuppositions may be avoided in therapeutic discourse, or that it may
have a clear pragmatic goal. Spenader (2002:116f) does not offer a clear answer to the question why the speakers in her
corpus use this strategy so frequently, but the heterogeneity of spoken discourse makes this problem hard to unravel.
The following section will investigate whether the reasons for intentional introduction of new information in
therapeutic discourse are easier to interpret and how often this strategy is used. A full-fledged comparison between the
strategies used individually by the therapist and the patient is beyond the scope of the present study, yet certain
differences will be illustrated.

Three factors will be considered in the analysis of the examples: (a) length and informativeness of the sequence
usually signalling that the information evoked is new (Prince, 1978), (b) the role of the message in the current
discourse (i.e. hearer-new information should be the speaker’s main point) (c) the hearer’s reaction (see Spenader,
2002:111). In more complex cases also the mechanism of accommodation will be described in closer detail. Consider
the first example:
(13) definite description, TP: there are other men (in Jennifer’s life) (S1: 190-195)
L: Well, in the first number of years we were in a stage that I think was fairly different.
We were younger and I think she was a different person, and I was a different person in the sense that our values
were different and our internal whatever drives us, I think, was slightly different. Things have to happen to us since
M. Andersson / Journal of Pragmatics 41 (2009) 721–737730
then, things that happen to everyone. But in those years, I would express a lot of emotion: ‘‘Why are you going out with
those other men?’’
Even though the expression
those other men is not very long and not particularly informative, it certainly constitutes
discourse-new information, as there are no traces of any ‘other men’ involved in this relationship in the entire session.
Furthermore, from the vantage point of the therapist the information is hearer-new, as he could have not known about
such details before (this is an initial session). However, Jennifer, who is aware of her own past, is also present in the
room and Larry’s planning of what should be said obviously builds mainly around this fact. Hence his utterance should
not be considered as intentional signalling of new information via presupposition, but rather sloppiness and lack of
precision. Yet the therapist is not in a particularly difficult situation when interpreting this sentence, as it rather clearly
shows the tension between what is said and what is generally established in people’s awareness. The latter is related to
the mainstream conviction that couples go to therapy when having problems and the problems can be other people
involved in their relationship. Larry is apparently aware of this fact, as he continues with his story in a matter-of-fact
mode afterwards and does not dwell on it. This confirms the view of Asher and Lascarides (1998) that not only the
presuppositions used to activate hearer-known information can constitute a background to support other new
information, but also these that introduce hearer-new information. In (13) the presupposition triggered by the nominal
expression
those other men supports the message concerning Larry’s emotions and the way he expressed them in the
situation he refers to. Such an interpretation is, apparently, acquired by the therapist also, since he never returns to this
subject. Consider another example introducing information that may be considered hearer-new:

(14) aspectual verb (a); TP: he has not thought about Janet before (S2:23–28)
definite description (b); TP: a woman exists
R: I guess what I have been trying to say is that women don’t like me.
T: Which woman, specifically?
R: Most women I meet.
T: Which woman, specifically?
R: Well, most women really but as you said that,
I just started (a) to think about this
one woman (b) – Janet.
T: Who’s Janet? (c)
As the above passage indicates, the patient expands a specific hurting experience and makes generalizations
about the entire female population. The therapist’s goal is to help him to deconstruct this preconception and
specify referents, which the patient apparently yields to and eventually picks out one. Since the therapist does not
object to this shift in Ralph’s point of view, we can conclude that the aspectual verb (
start) triggering the
presupposition in (a) is felicitously accommodated. Furthermore, that change of perspective was apparently the
therapist’s goal.
Yet the patient’s whole utterance is poorly planned, which the presupposition triggered by the definite
description (
this one woman) in (b) illustrates. The demonstrative here points to a referent that is specific only for
the patient, because the expression ‘this one woman’, without any clear antecedent, is almost equivalent to
indefinite ‘a woman’ (for discussion on specifics see Geurts (1999b). There is, certainly, nothing unusual in using
indefinite this-NPs to introduce a new discourse referent, however, as they are marked for definiteness, the hearer
needs to ‘repair’ the speaker’s discourse and accommodate a new entity. This is not always a straightforward
process, which the therapist’s answer proves: the patient caused an apparent obstruction in the communication and
therefore more information about the referent is requested, even though the name of the woman is immediately
spelled out afterwards.
12
Abstract triggers are accepted more easily than demonstrative NPs, which belong to the group of triggers that have
lower capacity to accommodate. This is probably related to the descriptive content of the former expressions

(Spenader, 2002:117; van der Sandt, 1992). Next example illustrates one of such cases:
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 731
12
Proper names also trigger presuppositions about certain individuals and are usually accommodated. Hence the name itself could not have been
sufficient for the therapist to recognise the referent unknown to him. However, the issue of proper names and their resolution is very complex and
cannot be an element of such a short study. For more detailed discussions see Kripke (1980), van der Sandt (1992), Geurts (1999a).
(15) aspectual verb; TP: he has watched Formula 1 before (S3:55–56)
P: I try to talk to him, but it’s always the same, I just talk and talk, carry on with my monologues and he
doesn’t even
stop watching Formula 1 ( )
The presupposition triggered by the aspectual verb (
stop) is certainly the main point of the patient trying to convey
the lack of communication between her and her husband and also part of a longer, informative utterance. Thus it can be
regarded as hearer-new information. In this case, however, the speaker’s ‘mistake’ is apparently easier to repair, since
aspectual verbs are semantically rich enough to evoke and foreground a change of state, not its initial stage, which is
taken for granted and therefore easily accepted (Geurts, 1999a:31). Yet both in Session 1 and Session 2 aspectuals
often bind to an antecedent in the preceding discourse, which means that they can be used more efficiently and refer to
previously discussed subjects. In (15) however, the speaker again seems to have poorly planned her utterance relaying
merely on the informative descriptive content of the presupposition trigger.
Consider an example in which the speaker seems to be aware that the conveyed information can be harder to
comprehend for the listener:
(16) factive; TP: those stupid jokes about guys make her laugh and think (S3:32–36)
P: ( ) one day I just
realized (a) that (b) those stupid jokes about guys, you know, like: ‘‘How can you impress a
guy? Get undressed and bring some beer’’, I (c) realized that such stupid jokes, sure, they still make me laugh,
but they also make me think ( )
T: It seems that you bear a grudge against men in general, or ?
The presupposition triggered by the factive verb (
realize) is a focal point of the patient’s story, which is typical
of hearer-new information (Spenader, 2002:112-ff; Spenader, 2003). Furthermore, the length of the whole passage

points to informativeness, a very important factor when introducing this kind of information. The presupposition in
(b) (
those stupid jokes about guys) has no antecedent in the previous discourse either and also has to be
accommodated; however, it is never problematic in the case of semantically rich big definite NPs.
13
Thus the
descriptive content of this presupposition on a par with the successive sentences contributes to the very long and
informative complement of the factive verb, which certainly eases the transition of knowledge between
interlocutors (additionally reinforcedbytherepetitionofthefactive
realize in (c)). This passage indicates that
there can be equilibrium between people’s discourse strategies, which is, unlike the preceding examples ((13), (14),
(15), (16)) more compatible with Spenader’s findings according to which the speaker may deliberately use
presupposition introducing hearer-new information.
Issues discussed in a therapeutic session are consensually validated by the patients trying to be as informative and
verbally efficient as they can and the therapists introducing alternative descriptions as smoothly as possible (Labov and
Fanshel, 1977:32; Buttny, 2001; Sluzki, 1975). There must be a certain feeling of reciprocity included in this exchange
on the part of the patient, as the solutions s/he expects to get can be offered provided that problems will be
comprehensibly described. We can, therefore assume that the therapist needs to have a special reason to deliver new
information as if it was a part of the common background (see (17) below), while such a situation in the case of the
patient is, probably, either a corollary of his/her verbal clumsiness and inefficiency or the result of being the constantly
challenged centre of attention. This is particularly visible in (14) where the patient apparently tries to avoid being
specific, while verbal inefficiency can be assigned to Larry’s utterance in (13). The following example illustrates how
presupposition can be used deliberately to introduce hearer-new information:
(17) definite descr iption; TP: somebody has life without him (S3:47–48)
T: But what will
your life without him look like, I mean after divorce?
P: Well I haven’t told you that we’ll get divorced
M. Andersson / Journal of Pragmatics 41 (2009) 721–737732
13
NPs modified by adjectives, relative clauses and prepositional phrases. See Spenader (2002:118).

The above passage illustrates probably the most prominent example of a communicational obstruction in all three
sessions. Even though possessive pronouns and big NPs accommodate very easily (van der Sandt, 1992), which
involves felicitous introduction of new information, apparently this view is too ‘technical’, as it accounts only for the
process of revising an incoming representation. Certainly, this process takes place in this case also, as it is not unusual
that married couples (particularly those that have problems) part and get separate lives without each other, so the
patient certainly comprehends the message. Yet the most important aspect of this context concerns her reaction that
clearly illustrates the cases in which the status of shared knowledge is disputable. In this case life without her husband
is for the patient merely part of general knowledge, not knowledge she wants to associate with her relationship.
However, provoking a potential discussion is probably part of the therapeutic strategy here. The entire tactic is
additionally reinforced by the temporal connective ‘after’, commonly regarded as an expression that triggers a
presupposition that something has happened (or here: will happen) (see Levinson, 1983; Geurts, 1999a,b; Schilder,
2001). The patient reacts as (probably) expected, however, considering the character of this reaction we can conclude
that felicitous accommodation of a new discourse element does not have to be tantamount to accepting it.
Presuppositional information can be denied, just as all information in a discourse can. This means that it will be
removed from the discourse context,
14
which the patient’s response illustrates.
Even though introducing hearer-new information via presuppositions can be quite an efficient strategy, the
therapists in the analyzed sessions do not resort to this means particularly often (2 times in Session 1 and 1 time in
Session 3). In Session 1 the therapist uses easily accommodating big NPs (here accompanied by a clausal complement)
to foreground his main point:
(18) definite description; TP: there are some fanatics (S1:133–135)
T: It sounds like you (to Jennifer) are on of
those fanatics that jump onto any bandwagon with all her soul, and
you (to Larry) say ‘‘Hey, wait a minute, reality testing, one, two, three.’’ (Jennifer and Larry laugh).
Since the patients do not object and react with laughter, we can conclude that the information was felicitously
accommodated. Furthermore, due to its more figurative than factual character the whole utterance could have been
treated as a type of humoristic retelling of the patients’ problems and not as hearer-new information par excellence.
The differences in use of presuppositions between therapists and their patients are rather conspicuous when we
compare (18) with, for instance, (13) or (14).

The therapists more often use presuppositions that accommodate in order to introduce discourse-new information.
Consider an example that illustrates the difference between hearer-new and discourse-new information:
(19) factive; TP: it is easy to say something (S1:251–253)
T: A technique’’ of getting in touch with his emotions may be for you (to Jennifer) to just listen and not react.
I
know that it’s easy for me to say. I’m an outsider. I’m not involved in an intense relationship like the two of
you are. But that would be, in my view, an important way of doing something for each other.
The presupposition triggered by the factive verb (
know) should be treated rather as discourse-new information, as
the it expresses an opinion that can be accepted on the basis of Stalnaker’s (1974) presumed common knowledge. We
cannot assume that the patient did not know before that advice is very easily given by ‘outsiders’, particularly in the
emotional sphere. Conversely, we can argue that this is a rather common belief, thus the reason why the therapist
asserts this knowledge might be a justification where a responsibility for the following utterances is evoked and, at the
same time, their non-offensive character emphasised (Buttny, 1993:16). From this vantage point the presupposition
evidently supports the information contributed by the whole sequence. This seems to be a very sophisticated technique,
as the process of accommodation involves revision of the incoming discourse representation structure with new
information by the hearer. The information, even though evoking rather bad news (it is easier to say than to do), yet
channelled through presupposition that cleverly re-describes and tallies with the patient’s own beliefs, seems to be easy
to accept.
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 733
14
For a more detailed discussion on denials in DRT (see Maier and van der Sandt, 2003).
Section 2.3 shows that Spenader’s (2002) findings about the intentional use of presuppositions in order to signal
hearer-new information in casual conversation only partly apply in the analyzed material, which implies a difference
between the genres. In her material in 58% (47 cases) of the cases the speakers seemed to believe that the presupposed
information is new to the hearer, while in the present study only 10.6% (15 cases) of the analyzed instances was
subsumed under this category. This divergence will be discussed in more detail in section 3 below, but it can already be
concluded that the strategy of introducing hearer-new information by means of presuppositions may be, in general,
avoided in therapeutic discourse, unless there is a specific reason for that. In the case of the patients this strategy is
probably the result of poor planning or verbal sloppiness (e.g. (13), (14)), while the therapists seem to use it more

consciously and more carefully ((17) or (18)). This observation points not only to different strategies apparently
deployed by the therapist and their patients, but also to differences between individuals and their personal capabilities
to furnish ideas, explain, explicate, and recapitulate them, all of which play a fundamental role in distribution and
efficacy of presuppositional information.
3. Discussion and conclusions
The above analysis of presuppositions in therapeutic discourse supports the opening prediction that binding may be
the predominant strategy for presupposition resolution in this genre. However, since presuppositions have not been
studied before in the context of therapeutic talk, the results of the present study are a first approximation and can hardly
be considered to be final. More corpus material needs to be examined in order to confirm or refute what has been said
here. Yet the question is what contributes to these early findings and, at the same time, to the discrepancies between
therapeutic discourse and casual conversation observed here?
The first reason seems to be the limitation of topics discussed in a therapeutic session, in which features once
examined are repeatedly referred to. This explains the high scores of bound NPs of each kind (in comparison to the
other triggers), particularly the demonstrative and possessive ones. A second reason is, as Davis (1986) writes, the fact
that among the most crucial elements of a therapeutic session are (re)formulations of the patients’ problems, which not
only provide interpretations but also are diagnostic. Therapists can use them in different ways, which is particularly
discernible in Session 2, where the therapist very actively participates in reformulating the patient’s generalized
version of his problems by making recurrent references to previously discussed issues via bound presuppositions.
Another reason for the predominance of bound presuppositions in therapeutic discourse may be the necessity of
mediating between the typical inaccuracy of the patient’s account and the therapist’s aim to achieve more specific
formulations. This is, again, particularly conspicuous in Session 2. Moreover, in this session the therapist confronts the
client via constantly used presuppositions that include mostly bound factive verbs. These verbs presuppose the truth of
their complements, which implies not just a harmless strategy of requesting information, but a sophisticated way of
challenging the patient’s generalized model of his/her situation.
Telling patients about their problems and making them aware of certain facts can be a delicate enterprise, as such
accounts do not always tally with their own beliefs. This can be the reason why in Session 1 presuppositional
expressions other than definite NPs (mostly bound to an antecedent) are rather scanty (Table 3) and the therapist is very
careful in making them. As Sluzki (1990:107) writes, the therapist’s task to maintain noncommitment in the case of
couple therapy is very complex, as one of the partners may feel alienated as soon as she or he can discern any overtones
of a ‘coalition’ with the other part. Indeed, statements that presuppose the truth of certain facts (e.g. factive verbs)

should be rather carefully distributed, particularly when they are part of multi-speaker discourse where they can be
much more easily verified and objected to. This can explicate why also the patients in Session 1 hardly use
presuppositional expressions other than definite NPs. It seems that actual states of affairs are very significant in couple
therapy. Hence even bound presuppositions that can be used rhetorically or objectify loose ideas (see (8)) may be more
difficult to make.
The above issues are not the case in Session 3 where the patient freely makes presuppositions that concern
subjectively perceived facts. The therapist’s task in this talk is often to soften them, as in (16) where she marks the
stance of her statement by using the doubt verb: ‘seem’ (on stance markers see Biber and Finnegan, 1989:120).
However, the most interesting feature in this session is the high scores for both resolution strategies (Table 3). The
difference between binding and accommodation in this particular session is not statistically significant, thus the figures
may have been obtained due to chance. However, since Session 3 is, unlike the two other talks analyzed, a subsequent
interview, therefore the question arises whether presuppositions are used differently in subsequent and initial
M. Andersson / Journal of Pragmatics 41 (2009) 721–737734
meetings. And, as Labov and Fanshel (1977:86) write, most information in therapeutic discourse is to be found not in
utterances themselves, but in the structure of shared knowledge. The foundations of such knowledge are, apparently,
stronger between those who know each other well than between people meeting for the first time. Thus assumptions
about common background made by the speaker can affect the form of linguistic production and, by default,
presuppositions. Yet this subject deserves a separate study where other corpora would be examined in order to confirm
or disprove this hypothesis.
The aforementioned claim by Labov and Fanshel (1977:86) is in conflict with Spenader’s findings about casual
conversation according to which most of the cases of accommodation occur when the speaker deliberately uses
presupposition to communicate information that is not based on shared or general knowledge.
15
However, in the
samples of therapeutic discourse analyzed, presuppositions used in this way are, most likely, a corollary of poor
planning or verbal inefficiency of the speaker, as, for example, in (13) and (14). Moreover, as many as 5 out of 11
presuppositions identified as introducing hearer-new information were triggered by an aspectual verb. Semantically
such verbs refer to the temporal structure of their complement verbs and serve as markers of aspect (i.e. beginning,
continuation, termination) of the eventuality expressed by these complements (Dowty, 1986). In the case of (15) the
verb ‘stop’ points to the end of the eventuality described, while the beginning and other stages of the event are

backgrounded and taken for granted. The foregrounded information is semantically rich and enables the
accommodation of this particular instance (and the whole group) without the danger of the misinterpretation on the
part of the hearer. In fact, the speaker is probably not even aware of using a presupposition when introducing hearer-
new information by means of aspectual or factive verbs, exactly due to their rich descriptive content. This gives a
sound explanation to the high scores of aspectuals used in this way and accommodated aspectuals in general.
In general, however, it seems that presuppositions as the means of introducing hearer-new information are used less
often in therapeutic discourse than in casual conversation (10.6% of the instances in the former genre and 58% in the
latter). The reasons for this are not hard to unravel. There is apparently not a big need to convey hearer-new
information in therapy talks that revolve around limited and recurrent topics. Furthermore, if patients try to be
informative and provide a comprehensive account of their problems, they may often choose other means to
communicate new information such as assertions. This is how a therapeutic session differs from a daily-life
conversation within which ‘hot news’ is ubiquitous and freely introduced. In the therapy it may be avoided as a source
of potential snarl-ups in communication. Thus the present paper seems to validate the classic view that presuppositions
are not often deliberately used when introducing hearer-new information.
The predominance of binding over accommodation in the analyzed samples of therapeutic discourse leads to one
more conclusion. The environment for the presentation of information in this genre is to a high extent pre-planned,
which concerns not only the aim-and-agenda-orientated therapist, but also the patient who wants to achieve certain
goals. This explains the big disproportion between abstract triggers in the two genres compared (see Table 2), as in the
therapy situation the speech of both interlocutors should be better organized and efficacy-oriented than in casual
conversation. To achieve such ‘tidiness’ the speaker needs to make use of certain rhetorical strategies, which suggests
the aptness of using bound presuppositions in cases when anaphors or other expressions could be used instead. One of
the aims of this paper has been to present the role and significance of such uses. Following van der Sandt’s (1992:360)
reasoning, presuppositions lose their presuppositional character if bound to a discourse-given antecedent and become
part of asserted information that is already present in discourse. Indeed, as the analyzed examples of binding indicate,
this resolution contributes to the cohesion and smooth flow of the discussion and efficiency in communication. Bound
presuppositions are, therefore, an appropriate ingredient of therapeutic discourse and can be very efficiently used
instead of their semantically simple anaphoric counterparts in this genre.
According to van der Sandt’s theory, binding is preferable as a way of resolving presuppositions. Yet, as Spenader’s
(2002) study indicates, naturally produced examples indicate that accommodation predominates and is felicitously
used in casual conversation (Tables 1 and 2). The fact that the participants in therapeutic discourse make utterances

that adhere to van der Sandt’s view and generate presuppositions that are more complementary with assertions
suggests that this theory may be, to some degree, wishful thinking about human speech. Therapeutic session takes
place in particular settings, and has specified aims and well-defined internal structure (agenda), all of which strongly
influence the reasons why certain utterances are generated. As the above reasoning indicates, for these reasons the
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 735
15
This claim concerns mostly factives and it-clefts, but, as Spenader (2002:116) argues, this seems to be a frequent use of accommodated
presuppositions.
number and distribution of presuppositions made in therapeutic discourse may differ from those made in an ordinary
conversation.
In summary, the study shows that the strategies for the presupposition resolution proposed in the binding theory can
be felicitously used both in therapeutic interaction and in casual conversation. Furthermore, the roles they play in the
interaction, the ways in which they facilitate the communication (e.g. when used instead of semantically simpler
anaphors) or obstruct the communication are very similar in both genres. What is different are the underlying reasons
for making presuppositions and the effect they can have on the possible changes in the patients’ thinking.
4. Future work
Certainly further investigation of empirical material is needed in order to verify the initial observation that binding
may be a predominant strategy for presupposition resolution in therapeutic discourse, unlike in casual conversation.
Also possible differences between initial and subsequent sessions are worth further examination in order to judge if
they constitute an invariable characteristic of this genre. More empirical data should also be analyzed to determine
more satisfactorily if the avoidance of introducing hearer-new information via presuppositions indicated in the present
paper is constant in therapeutic talks. Moreover, there are a lot of different therapeutic strategies and it would be
illuminating to investigate whether the method acquired by the therapist influences both their own presuppositions and
those made by patients.
Another interesting idea worth scrutinizing in therapeutic discourse would be presuppositions triggered in
intensional contexts, i.e. modals and attitude reports, which may be rather commonly used in this genre, as they have
obvious emotive functions, particularly important when expressing feelings and emotions.
Acknowledgments
I want to thank Jennifer Spenader for giving me the most useful advice and commenting on the earlier versions of
this study. Many thanks also go to Philip Shaw and Nils-Lennart Johannesson for giving me very valuable insights

about the way of formulating ideas. All errors are my own.
References
Ajimer, K., 1996. I think—an English modal particle. In: Swan, T., Westvik, O.J. (Eds.), Modality in Germanic Languages. Historical and
Comparative Perspectives. Mouton de Gruter, Berlin, New York, pp. 1–48.
Asher, Nicholas, Lascarides, Alex, 1998. The semantics and pragmatics of presuppositions. Journal of Semantics 15 (3), 239–300.
Bandler, Richard, Grinder, John, 1975. The structure of magic. In: A Book About Language and Therapy, Science and Behaviour Books, Inc, Santa
Clara.
Biber, Douglas, Finnegan, Ruth, 1989. Styles of stance in English: lexical and grammatical marking of evidentiality and affect. Text 9 (1), 93–124.
Blanchet, Alain, Batt, Martine, Trognon, Alain, Masse, Laurence, 2005. Language and behaviour patterns in a therapeutic interaction sequence. In:
Anolli, L., Duncan, Jr., S., Magnusson, M.S., Riva, G. (Eds.), The Hidden Structure of Interaction: From Neurons to Culture Patterns.
IOS Press, Amsterdam, pp. 124–140.
Buttny, Richard, 1993. Social Accountability in Communication. Sage Publications, London, New Dehli.
Buttny, Richard, 2001. Therapeutic humor in retelling the clients’ tellings. Text 21 (3), 303–326.
Clark, Herbert, 1975. Bridging. In: Schank, R., Nash-Webber, B. (Eds.), Theoretical Issues in Natural Language Processing. MIT, Cambridge.
Reprint from: Johnson-Laird, P.N, Wason, P.C. (Eds.), 1983. Thinking. Cambridge University Press, pp. 411–420.
Davis, Kate, 1986. The process of problem (re)formulations in psychotherapy. Sociology of Health and Illness 8, 44–74.
Delin, Judy, 1995. Presuppositions and shared knowledge in it-clefts. Language and Cognitive Processes 10, 97–120.
Dowty, David R., 1986. The effects of aspectual class on the temporal structure of discourse: semantics or pragmatics? Linguistics and Philosophy 9,
37–61.
Eggins, Susanne, Slade, Diane, 1997. Casual Conversation. Cassel, London, Washington.
Fraurud, Kari, 1990. Definiteness and the processing of NP’s in natural discourse. Journal of Semantics 7, 395–433.
Gazdar, Gerald, 1979. Pragmatics, Implicature, Presupposition and Logical form. Academic Press, New York.
Geurts, Bart, 1999a. Presuppositions and Pronouns. Current Research in the Semantics/Pragmatics interface, vol. 3. Elsevier Science Ltd,
Amsterdam, New York, Oxford, Tokyo.
Geurts, Bart, 1999b. Specifics. Retrieved from: />Halliday, Michael A.K., Hasan, Ruquaya, 1976. Cohesion in English. Longman, London.
Hilsenroth, Mark J., Cromer, Thomas D., 2007. Clinician interventions related to alliance during the initial interview and psychological assessment.
Psychotherapy: Theory, Research, Practice, Training 44 (2), 205–218.
M. Andersson / Journal of Pragmatics 41 (2009) 721–737736
Kamp, Hans, 1981. A theory of truth and semantic representation. In: Groenendijk, J.A.G., Stokhof, M.B.J. (Eds.), Formal Methods in the Study of
Language. Mathematisch Centrum, Amsterdam, pp. 277–322.

Kamp, Hans, Reyle, Uwe, 1993. From Discourse to Logic. Kluwer Academic Publishers, Dordrecht.
Karttunen, Lauri, 1973. Presuppositions of compound sentences. Linguistic Inquiry 4 (2), 167–193.
Karttunen, Lauri, 1974. Presupposition and linguistic context. Theoretical Linguistics 1 (1), 181–194.
Kiparsky, Carol, Kiparsky, Paul, 1971. Fact. In: Steinberg, D.D., Jakobovits, L.A. (Eds.), Semantics: An Interdisciplinary Reader in Philosophy,
Linguistics and Psychology. Cambridge University Press, Cambridge.
Kripke, Saul A., 1980. Naming and Necessity. Blackwell, Oxford.
Labov, Wiliam, Fanshel, David, 1977. Therapeutic Discourse. Academic Press, New York, London.
Lambrecht, Knud, 1994. Information Structure and Sentence Form: Topic, Focus, and the Mental Representations of Discourse Referents.
Cambridge University Press, Cambridge.
Levinson, Stephen C., 1983. Pragmatics. Cambridge University Press, Cambridge, London.
Maier, Emar, van der Sandt, Rob A., 2003. Denials in discourse. Retrived from: />Prince, Edward, 1978. A comparison of Wh-clefts and it-clefts in discourse. Language 54, 883–906.
van der Sandt, Rob A., 1992. Presupposition projection as anaphora resolution. Journal of Semantics 9, 333–377.
Sbisa
`
, Marina, 1999. Ideology and the persuasive use of presuppositions. In: Verschueren, J. (Ed.), Language and Ideology. Selected Papers from the
Sixth International Pragmatics Conference, vol. 1. Internatonal Pragmatics Association, Antwerp, pp. 492–509.
Schilder, Frank, 2001. Presuppositions triggered by temporal connectiveness. Semantics and Pragmatics of Discourse and Dialogue: Experimenting
with Current Theories, vol. 9. Elsevier Science Ltd., Amsterdam, New York, Oxford, Tokyo, pp. 27–43.
Sluzki, Carlos E., 1975. The Coalitionary Process in Initiating Family Therapy. Family Process 14 (1), 67–77.
Sluzki, Carlos E., 1990. Systemic blueprints in a therapeutic conversation. In: Chasin, R., Grunebaum, H., Herzig, M. (Eds.), One Couple, Four
Realities: Multiple Perspectives in Couple Therapy. Guilford, New York, pp. 107–125.
Spenader, Jennifer, 2002. Presuppositions in spoken discourse. Doctoral Dissertation, Department of Linguistics, Stockholm University.
Spenader, Jennifer, 2003. Factive presuppositions, accommodation and information structure. Journal of Logic Language and Information 12, 369–
374.
Stalnaker, Robert, 1974. Pragmatic presuppositions. In: Munitz, M.K., Unger (Eds.) Semantics and Philosophy, New York, pp. 1972–213. Reprinted
in: Davis, S. (Ed.), 1991. Pragmatics. Oxford University Press, New York, Oxford, pp. 471–485.
Marta Andersson is currently a PhD candidate at Stockholm University. Her interests comprise discourse analysis, text linguistics and also issues
concerning language competence and use from a semantic, pragmatic and psycholinguistic perspective.
M. Andersson / Journal of Pragmatics 41 (2009) 721–737 737