Documente Academic
Documente Profesional
Documente Cultură
On mitigation *
Claudia Caffi
Department of Scienze Glotto-Etnologiche, University of Genoa, Via Balbi, 1-16126 Genoa, Italy
"Vostra Signoria mi scriva quali parole pare a lei che
debbano esser mitigate, ch'io mi sforzerb di mitigarle."
Torquato Tasso, Lettera del 14 di Maggio 1575, in: Le lettere di Torquato Tasso, disposte per ordine di tempo e illustrate da Cesare Guasti, Firenze, Le Monnier, 1852-1855,
vol. I, p. 75.
Abstract
I wish to thank the organizers of the Third Rasmus Rask Colloquium, Odense, 5-7 November 1996,
Hemming G. Andersen, Leo Hoye and Johannes Wagner for their kind invitation and hospitality. I am
indebted to Shoshana Blum-Kulka, Alessandra Fasulo, Klaus H61ker, Richard W. Janney, Clotilde Pontecorvo, and Marina Sbis~tfor insightful comments on an earlier draft. I'd like to thank Ian Harvey for
his careful stylistic revision. I'm also grateful to the anonymous doctors and patients who were not afraid
of the recording, thereby making possible the research partly reported here. This paper - which summarizes some issues I deal with extensively in my book on mitigation (in preparation) - would not have
been possible without Jacob Mey's work in pragmatics.
0378-2166/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved.
PII: S0378-2166(98)00098-8
882
I. Introduction
This paper aims at recasting the issue of mitigation in a broad, integrated pragmatic framework by bridging different approaches to the study of communication, in
particular, rhetorical and psychological approaches, which tend to remain too distant
from each other and are only occasionally linked. The general view of communication I subscribe to is that of a complex system where many parameters - be they discrete or scalar, central or peripheral - interact and can be labeled as systemic, holistic. I will use the term 'parameter' in more or less its everyday sense of 'a quantity
whose value varies with the circumstances of its application' (Webster's new worm
dictionary, 1988). The issue of the hierarchical organization of parameters and their
relationship in given contexts, e.g. their mutual implication, is beyond the purposes
of the present paper. However, the basic reason for adopting this term and its systemic theoretical framework is summarized by one of the premises of system theory
applied to human communication, namely the assumption that the parameters of an
open system are more important than the system's initial conditions (Watzlawick et
al., 1967: 4.33). The advantage of this perspective lies in the fact that it allows a
dynamic approach to interaction where the focus is on the system's functioning and
contextual organization.
According to my definition, mitigation - which I take to be a synonym for attenuation - is the result of a weakening of one of the interactional parameters, and a
downgrading when the parameters involved are scalar. Mitigation is one of the two
directions of modulation,1 namely the rhetorical stylistic encoding of an utterance
(Caffi, 1991), its expressivity, opposed and complementary to the direction 'reinforcement'. Hence, it is a superordinate organizing concept to which it is possible to
ascribe different functions performed by heterogeneous linguistic means that are
labeled variously in pragmatic research (e.g. Edmondson, 1981; House and Kasper,
1981; Blum-Kulka et al., 1989; for a survey, see Kerbrat-Orecchioni, 1992). As a
consequence of the weakening of an interactional parameter, mitigation locally
affects the allocation and reshuffling of rights and duties triggered by the speech act,
and, crucially, changes their intensity and cogency. Globally, it reduces participants'
obligations (Meyer-Hermann and Weingarten, 1982: 243), to which the felicity conditions of a speech act belong, thereby furthering the achievement of interactional
goals. Thus, mitigation is functional to smooth interactional management in that it
reduces risks for participants at various levels, e.g. risks of self-contradiction,
refusal, losing face, conflict, and so forth. This definition enables us to explain why
mitigating devices in natural languages studied so far are much more numerous than
reinforcing devices. To give one example, with reference to requests in English and
1 The term 'modulation' is used in HaUiday (1976: 200) to refer to "a kind of quasi-modality" and in
Lyons (1977: 65) to refer to "the superimposing upon the utterance of a particular attitudinal colouring,
indicative of the speaker's involvement in what he is saying and his desire to impress or convince the
hearer". The term is also used in Georges Mounin's many works on translation with reference to one
technical procedure of translation. My different use of the term takes inspiration from physics, where
modulus means 'intensity', which is a scalar parameter defining a measurable quantity, e.g. a force.
883
German, House and Kasper (1981) list eleven types of downgraders and six types of
upgraders. The basic effect of reducing obligations makes it possible to unify mitigation which relates to deontic modality and mitigation which relates to epistemic
modality. Typically, mitigation affecting deontic modality reduces addressee's obligations, while mitigation affecting epistemic modality reduces speaker's obligations.
To use a distinction advanced in a different framework by Giles et al. (1979), I propose to gather the different functions of mitigation around two main dimensions,
which also differ in conventionality: (1) the dimension of interactional efficiency,
which meets essentially instrumental needs, since mitigation helps achieve interactional goals; (2) the dimension of identity construction, which meets essentially relational needs, since mitigation is functional to the monitoring of emotive distances
between interlocutors.
Given these starting points, I will argue that mitigation works in a multi-layered
and multi-dimensional way, simultaneously affecting a plurality of linguistic levels
and interactional dimensions. The negative character of the speech act for the
addressee, its being a face-threatening act (Brown and Levinson, 1978), with which
mitigation is usually associated, is just one of these parameters; and these can be
central or peripheral, according to the type of interaction and its goals. So, my first
aim is to show, tentatively, and rather ethnocentrically, as the examples are in Italian, how mitigation affects different parameters and different dimensions at the same
time. More precisely, I will try to show that mitigation is active in both the dimensions distinguished above, because it involves psychological and emotive aspects
which have so far been neglected in favor of sociolinguistic aspects (Blum-Kulka,
1992: 273). In particular, using examples of different types of mitigation from a corpus of doctor-patient interaction in Italian, I will try to illustrate the multi-level
process defined by Giv6n (1989), i.e. the inferential, abductive process whereby
grammatical and semantic phenomena shade into pragmatic modalities and psychological effects. I will focus on the contribution that mitigated choices can make to
the emotive monitoring of interaction, lending themselves easily to an increase or
decrease in psychological distance. This issue, which concerns the less conventionalized (and less studied) steps of the inferential process, is in my view a crucial one,
inasmuch as the management of psychological distance is part and parcel of our
communicative competence in both encoding and decoding processes (Frijda, 1982:
112). We are faced here with very subtle inferences which are easy to grasp and difficult to describe and which once again foreground the problem of the integration of
affect and emotive communication in pragmatics (Arndt and Janney, 1987; Caffi and
Janney, 1994). This research perspective has been partly anticipated by German
research on image-work (Imagearbeit) and relational work (Beziehungsarbeit) (Holly,
1979; Adamzik, 1984) which have developed Goffman's concept of 'face work'.
My second concern here will be to put forward - more as a heuristic tool than as
a cross-cut typology - a tripartite classification of mitigating devices based on different scopes of mitigation. I will advance a classification of mitigating mechanisms
based on the three components of the utterance on which mitigation can operate: the
proposition, the illocution, and the utterance source. Extending Lakoff's (1973)
metaphor, I'd like to call them 'bushes', 'hedges', and 'shields' respectively. The
884
formal criterion I have selected is clearly only one of the possible classificatory criteria. Since it is anchored in linguistic form, it has, however, the advantage that it
avoids the risk of an unfettered proliferation of maxims which vary according to the
linguist's sensibility and imagination (from 'be tactful' to 'be optimistic' etc.). Furthermore, I will distinguish some mitigating strategies which concern larger units
than the utterance.
My third point, which here will be presented only as a working hypothesis, takes
into account some possible links between scopes of mitigation and their potential
effects on the relational emotive level. In order to clarify these kinds of effects, I will
apply a non-linguistic model, i.e. the theory of disqualification, advanced in 1959 by
Jay Haley, one of the founders of family therapy, to explain the functioning of schizophrenic communication in a systemic framework. The pragmatic notion of 'mitigation', in its broad as well as in its narrow sense, shares with the psychological notion
of 'disqualification' the basic feature of deresponsabilization. Both notions call into
play the problem of responsibility management in discourse, involving cognitive as
well as emotive aspects.
885
886
ferent languages, has shown, though not emphasized, that apart from the set of specialized means typical of a given illocution, mitigating strategies are basically similar across illocutions. In line with these findings, I compared, in a previous stage of
my research on mitigation strategies in Italian (Caffi, 1995), the linguistic means
employed to mitigate different types of speech acts, which I provisionally subdivided
into two macro-types according to their speaker- or hearer-centeredness and their
'direction of fit' (Searle, 1975), i.e. constative-verdictive acts vs. directive-exercitive
acts. I expected to find different classes of mitigating devices reducing the speaker's
obligations in the case of constatives-verdictives, and the hearer's obligations in the
case of directives. As it happened, it emerged instead that, from a formal viewpoint,
downgraders in both illocutionary macro-types can be clustered around two types,
which can be labeled 'substitutive' and 'additive' (cf. Lausberg, 1967; KerbratOrecchioni, 1992: 200). Substitutive means are at work in so-called 'indirect speech
acts' (Searle, 1975) which question or state felicity conditions of a speech act, e.g. in
requests, preparatory conditions for the performance of the action. Another kind of
substitutive means is represented by 'shields', which will be addressed under s. 4.
Among the additive means, the following can be employed in many different illocutions. With regard to 'internal' mitigation (for the distinction between internal and
external mitigation, see Blum-Kulka et al., 1989): morphological means, e.g.
diminutive suffixes, vocative, address terms (the apostrophe in rhetoric); syntactic:
(a) local, e.g. conditional mood; (b) global, e.g. hypothetical constructions; lexical,
e.g. markers such as un po ', un attimo, magari. With regard to 'external' mitigation:
if-clauses, pre-sequences, grounders. Thus, my starting hypothesis about distributional constraints of mitigators according to the type of illocution proved to be somehow inadequate, since data showed rather a bi-partition of mitigators from a functional viewpoint into two basic groups: the group of 'illocution-free' mitigators and
the group of 'illocution-bound' mitigators, specialized for a specific illocution. In
particular, specialized mitigators cover the whole range of epistemic mitigators specialized for statements, the 'evidentials', from modal adverbs to parenthetical verbs,
which are simultaneously indicators of mitigation and indicators of illocution. As it
has been pointed out by previous research on modals, probabilmente ('probably'),
for example, not only downgrades the illocutionary force by downgrading the epistemic commitment to the proposition, but also indexes the act as a statement. Similarly, among the whole range of specialized mitigators for requests, per favore
('please') downgrades the directive and at the same time indexes the act as a request.
On the other side, there are downgraders like un attimo, un po', magari and other
passe-partout downgraders that can be used in various illocutionary acts (where they
modify various aspects). There is an inclusive relationship between the two sets of
mitigation; in other words, the set of mitigation markers includes a subset of specialized markers which are also markers of illocution. The specialization process
could be explained diachronically by the conventionalization of devices, at different
degrees of transparency, which progressively lose their semantic content and become
'frozen' mitigators (Labov and Fanshel, 1977: 83). Theoretically, a reasonable
hypothesis to explain the process of specialization of some mitigators into illocution
markers, which could be studied as cases of grammaticalization, is that these latter
887
markers are directly connected with the essential condition, in Searle's sense, of the
act (the idea of linking mitigation and felicity conditions is proposed in Lakoff,
1980). Instead, mitigators like un attimo, un po', p e r caso, magari etc. work by
implicature and are based on very general semantic mechanisms, such as temporal or
logical operations (see below 5.4.).
888
fessional reliability) enhances the degree of conversational cooperation and the number of inferential abductive steps on both sides (Arndt and Janney, 1987). We have
no direct access to the speaker's mind, and the best we can offer is a description of
what we are entitled to infer on the basis of actual words, a reconstruction of recognizable effects which is compatible with a given context. Thus, what is crucial from
this perspective is not individual psychology, but rather the links between interdependent variables (psychological, sociological, linguistic variables) within the interactional system. A pragmatic theory must account for all these variables together,
not one after another, piece after piece, as if communication were a machine) A general systemic approach makes it possible to ask questions such as: how can surface
stylistic choices contribute to building up the interpersonal relationship? And this
question forms part of a set of more general questions related to the definition of the
'subjectivity': who is speaking in pragmatics? Who are the inhabitants of pragmatics as a general theory of human interaction? Or, to rephrase the title of one of Jacob
Mey's books (Mey, 1985), whose pragmatics? Is the model speaker of pragmatics,
like Winnicott's (1965) false-self, merely conforming to external social expectations
(a definition of 'face' may be 'just what society expects us to conform to')? Today's
pragmatics, having overcome the phase in which the anthropological, historical, psychological subject had to be suppressed in favor of his/her formal simulacrum, is
now allowed to raise these questions.
4. Scopes of mitigation
By way of starting a systematic account, I will concentrate on the speech act unit.
For the sake of space and clarity, I will exclude cases of sequential or multi-turn mitigation, in particular, the case of passing on bad news, and phenomena such as the
'perspective display', a kind of mitigation in the delivery of a diagnosis (Maynard,
1992). Leaving aside the external mitigation obtained by pre-sequences, grounders,
etc. (Blum-Kulka et al., 1989), different kinds of mitigation can be distinguished
according to their different scopes in the speech act, according to the abstract component on which mitigation centers. These abstract scopes can be gathered around
three major focuses: the proposition (and within it, the reference and the predication), the illocution (and within it, the speaker's propositional commitment toward
the proposition in assertive speech acts), and the deictic origin of the utterance (Btihler, 1934), the 'I-here-now', the ego-hic-nunc at the core of Benveniste's (1970)
instance d'~nonciation. Obviously, more than one type of mitigating device can be
employed simultaneously in each case, and conversely, one mitigating device can
mitigate more than one aspect of the speech act simultaneously.
Following and extending Lakoff's (1973) metaphor, I'd like to call these different
types of mitigators 'bushes' (i.e. propositional hedges), 'hedges', and 'shields'
respectively. There are no clear-cut distinctions between these cases, since the dif-
889
ferent parts of the speech act are not neatly separated. Furthermore, the three kinds
of mitigating devices are heterogeneous, inasmuch as bushes and hedges are lexicalized expressions, sometimes markers: if the latter is the case, it is often problematic
to assign them a scope, also due to the well-known fact that markers are often not
integrated syntactically into the sentence (disjuncts). Instead, in shields, unlike in
bushes and hedges, there is no explicit operator of mitigation and the weakening
operation takes place at a deeper, more abstract level: for instance, it affects syntax,
as in passive transformations, or morphology, as in the shift from first-person singular pronouns to other person pronouns.
Approximate as this triad may be, I find it useful in making some first heuristic
distinctions. And, crucially, my point of view is not t a x o n o m i c but pragmatic: my
goal is not to classify expressions but to explain occurrences, and my concern is not
with words as such but with their use (this point is treated extensively in H61ker,
1988).
5. Some examples
In this section, I will discuss some examples integrating different perspectives, in
order to show h o w mitigation works.
The following examples are taken from transcripts of doctor (D) - patient (P) and
psychotherapist (T) - client (C) conversations recorded from 1994 to 1996 in Northern Italy. 4 The English translations are only tentative. I have dealt with some pragmatic aspects of this type of asymmetrical interaction also in Caffi (1997). It is
impossible here to gloss the examples extensively; nevertheless, I have taken into
account the whole conversation. Nor is it possible to consider the very important
4 Abbreviations used are the following:
GP
general practice visit
SV
specialistvisit
PsV
psychiatricvisit
PsS
psychotherapeuticsession
Transcript conventions used are the following:
short pause (up to 2" duration)
-pause (between 2" and 5")
::
lengthenedsyllables
=
"latched" utterances
falling intonation contour
?
rising intonation contour
->
intonationcontour of "non-finality", level intonation or "holding intonation in which the
tone neither falls nor rises" (Gumperz 1992:235) (Italian "intonazione sospensiva")
*+
overlap (*marks the beginning in A's and B's turns;+ marks the end in A's and B's turns)
(h)
audible breath
XXX
additionalpitch
**
low volume
(italics) non-linguistic phenomena
xx
unintelligiblewords
((XXX)) omissions
890
891
toward regression. In a less pessimistic and more general version, diminutives "may
at large be seen as in-group markers that fall into the sphere of our notions of familiarity, intimacy and decreased psychological distance ... [they] may also function as
'accelerators' of intimacy" (Dressier and Merlini Barbaresi, 1994: 233). In the psycholinguistic framework developed by Wiener and Mehrabian (1968), the diminutive
in (1) could be described an as indicator of immediacy. Significantly, in my corpus
of 'therapeutic' interaction, the use of this morphological mitigating resource, which
also occurs in (2) and (3) below, is extremely frequent, sometimes becoming a distinctive feature of a doctor's communicative style that makes it very close to a sort
of baby-talk. If the benefits of mitigation sketched above are clear, both from an
instrumental and a relational viewpoint, it also implies potential costs: for instance,
the patient, feeling him/herself treated like a child, may reject this definition of the
relationship (saying something like 'not as bad as the last one you gave me') or may
continue the interaction on excessively intimate terms so that intimacy 'brakes'
would have to be deployed by the doctor.
Examples (2) and (3) are understatements in diagnosis, which is a verdictive
speech act in Austin's (1962) sense, where mitigation, centered on propositional
content again, affects the parameter of 'precision'.
(2) D. la sua non 6 una vera e propria emia. - ~ solo un pochino:=
P. =una puntina.
D. una puntina. (GP)
D. yours is not a real hernia. - just a bit =
P. a spot+DIM
D. a spot+DIM
The doctor, after his attempt at mitigating the diagnosis, by using an understatement
and a minimizer in the predication ('it is just a bit of a hernia'; un pochino is a
diminutive form of un po'), accepts the non-technical definition suggested by the
patient in his turn-completion. The doctor's mitigation is reformulated through the
patient's mitigation on which agreement is finally reached. Much more than a simple terminological agreement, this means a convergence of two codes, a typical case
of accommodation in Giles and Coupland's (1991) terms (the powerful member
adapts his/her code to the other's code). Further, the convergence not only concerns
definitions and codes (the technical and the non-technical) but also styles, reached
via mitigation: the patient volunteers a re-formulation of the diagnosis (una puntina)
which is both colloquial and mitigated with a bush (the diminutive suffix). Through
the joint sequential definition as well as through mitigation, different kinds of effects
are reached: first, on a purely referential level, the health problem is, though imprecisely, co-defined along a three-turn sequence as a non-prototypical case of a given
disease and therefore not particularly serious. On a relational level, not only the
social but also the psychological distance between the interlocutors (a general practitioner and a South Italian shoemaker in a big Northern town) is diminished. Certainly, the sequential format of the exchange plays an important role in producing
892
this effect: the co-production of the problem definition is an interactional achievement which contributes per se to building a co-operative climate. This co-production
also involves, however, stylistic aspects. The partners' sharing of the minimizing
attitude expressed in the mitigated informal style redefines their relationship and
puts it on a more friendly basis. This example, which shows the intertwining of
sequential and stylistic aspects of interaction, prompts a general remark: a pragmatic
account which separates, as Levinson (1992) does, structural and stylistic aspects is
at best partial, at worst misleading.
In the two-part turn of (3), mitigation is obtained through morphological means,
i.e. the diminutive suffix and the pronoun quello ('that'), a case of empathetic
deixis, which bears a feature of [--PROXIMITY] (in paradigmatic opposition to
questo, 'this', which should have been selected in the co-text, since the problem at
hand had just been introduced by the patient in the adjacent turn) conveying a negative attitude:
(3) D. ma quello 6 un problemino. - non 6 mica un problema grosso. (GP)
'but that's a problem+DIM. - it's not a big problem'
Mitigating syntactic means are used in the second part of the turn, i.e. a litotes
shaped as a negatio contrarii (non d mica un problema grosso), which reformulates
the propositional content of the first part with a stylistic variatio. From an instrumental viewpoint, these bushes minimize the seriousness of the problem, which in
any case remains undefined by the expert in precise terms. From a relational viewpoint, it could be observed that the reassuring function of these minimizations - the
benefit - works somehow to the detriment of the interlocutor's reliability: he is indirectly treated as excessively worried, a bit fussy, maybe an anxious subject. Indeed,
research on doctor-patient interaction has repeatedly shown that a typical doctor's
strategy is to treat requests for information as boring symptoms of anxiety.
5.2. Hedges
In hedges, the scope of the mitigation centers on the illocution, i.e. on illocutionary force indicators. My view of speech acts as clusters of multi-level and multidimensional features which range along a continuum of gradual variations both
within a given type of illocutionary force and between two different types (cf.
Giv6n, 1989; Bazzanella et al., 1991) leaves no space for a notion such as that of
'indirect speech act'. The latter would also be in contradiction with my claim (see
7 below) about the non-dichotomic functioning of bushes and hedges. This issue
deserves, however, a full discussion which for reasons of space I can't enter into
here.
In examples (4) and (5) we find hedges.
(4) T. io le proporrei se vuole una medicina apposta per vedere se riesco a farla
dormire.
P. hmm (PsV)
893
T. I'd propose, if you like, a special medicine, to see if I can make you sleep.
P. hmm
In (4), the mitigating devices are the following:
(1) io le proporrei, a hedged performative: the verb in the conditional mood is a
weaker form than the performative expression 'I propose'; this is a case of
internal mitigation according to Blum-Kulka et al.'s (1989) model.
(2) se vuole ('if you like'), a routine formula, a consultative device (Blum-Kulka et
al., 1989: 19).
(3) A supportive postponed move, a case of external mitigation, a grounder (of the
Head Act): a final clause, per vedere se riesco afarla dormire ('to see if I can
make you sleep') which, according to Blum-Kulka (1992: 267), applies the negative politeness principle by appealing to the hearer as a rational partner who
cannot be forced to do something that s/he does not fully understand.
From a relational perspective, on the one hand, the doctor presents the therapy as
an attempt whose possible success will be his personal merit, thereby indirectly
affirming his strong role, as the 'healer'; on the other hand, he formally downgrades
his directive to a proposal which it is up to the patient to accept. Besides, from this
relational, emotive perspective, there is an ambivalence in the doctor's utterance
about who is the main actor in the therapeutic process, with an oscillation between
two different 'narratological' answers to the question 'who is the protagonist?' - and
correlatively between two different cultural models of the professional's role.
The modal probabilmente occurs twice in the following example:
(5) D. probabilmente 6: - dove c'~ l'attacapanni - probabilmente 6 una:: conseguenza di un problema intestinale: che ~ cominciato con l'influenza eh:
(SV)
'probably it is - where the clothes-stand is - it is probably a consequence of
an intestinal problem: that began with the flu eh:'
Here, the scope of the mitigation is (that aspect of the illocution which is) the speaker's epistemic commitment to the propositional content. Probabilmente weakens the
speaker's degree of certainty about the proposition: the overall effect on the utterance is that the diagnosis is downgraded to a hypothesis, nothing more than an
attempt to trace abductively the temporal and causal sequence of facts. Pragmatically, the diagnosis can be analyzed as a verdictive speech act in Austin's (1962)
terms, a type of illocutionary act whose preparatory felicity conditions have to do
with the authority or the competence of the speaker-agent. For these reasons, diagnosis is easily able to show the interplay of different parameters - knowledge,
power, role - in the same speech act. The assignment of a minor value to the parameter of knowledge in (4) and (5) goes together with a downgrading of the parameter of power. The semantic indeterminacy associated with the litotes in (4) and the
downgrading of certainty on the part of the speaker in (5) are mirrored by the
894
upgrading of the hearer's active role, e.g. his/her involvement in deciding what the
case is, in confirming the hypothesis, in adducing further evidence etc., in a word, in
sharing the responsibility for the abductive process. From a relational perspective,
this means a redefinition of the role of the patient as slightly less passive. Further, if
the patient is also entitled to a kind of knowledge, the relationship may proceed on a
more symmetrical basis: at this point a range of possible emotive inferences opens
up. And this range includes the possible interpretation on the part of the patient that
the doctor's decreased epistemic subscription to the content may be paralleled by a
decrease in assertiveness, to use a psychological category (Arndt and Janney, 1987),
in what I'd like to call an 'emotive subscription', an identification of the speaker
with his/her communication. In fact, modal adverbs are listed as indicators of nonimmediacy, precisely in the 'modification' class, in Wiener and Mehrabian (1968:
44). In conclusion, what the examples show clearly is that epistemic certainty, social
power, and psychological distance are connected parameters in a multilayered
process which, especially in its less conventionalized steps, remains to be deconstructed.
5.3. Combinations of bushes and hedges
In the following examples, mitigation affects both the proposition and the illocution. In example (6), the parameter of 'epistemic certainty' is crucially affected by
mitigation and is assigned a minor value, while the parameter 'intimacy' or 'emotive
closeness' is indirectly assigned a major value. In particular, this example raises an
important theoretical question: what is the threshold beyond which mitigation produces not only a deintensification of the same act but a different act, i.e. when the
problem at hand is not a matter of degree but a matter of type (Giv6n, 1989)? In (6),
the weakening of the 'epistemic certainty' parameter affects the parameter of 'social
role', here a professional one, in the overlapping area of 'competence'. This weakening, obtained through a combination of bushes and hedges, is such that the speech
act seems to shift from one type to another. It becomes, more than a hypothesis, a
verdictive, the giving up of any hypothesis, the expression of a doubt, a behabitive,
the report of a mental state, which hopefully can be limited. This shift happens when
the sincerity condition (Austin's, 1962, F-conditions) prevails over the competenceknowledge condition (Austin's, 1962, A-conditions) crucial to verdictives.
(6) D. magari ~ un periodo cosl - va a sapere - qualcosa del genere. (GP)
'maybe it's a sort of bad moment - who knows - something like that'
In (6), magari/va a sapere are hedges focussed on the aspect of the assertive illocution which is the commitment to truth (the neustic); d u n periodo cosi/qualcosa del
genere are bushes that make the proposition semantically fuzzy.
On the relational level, in (6) the downgrading of the knowledge parameter (by the
lexical means magari; the syntactic bushes ~ un periodo cosi and qualcosa del
genere which make the reference fuzzy; and finally by the explicit admission of
uncertainty va a sapere), which is even stronger than in (5), increases the symmetry
895
from a social viewpoint and the (not necessarily welcome) intimacy from a psychological viewpoint. It is as if the doctor, by giving up his knowledge, also gives up his
social role based on that knowledge: both seem temporarily suspended.
(7) is another example of the combined use of bushes and hedges:
(7) T. niente mh sl - allora se ho capito bene - il suo problema ~ che ogni tanto le
capita di svenire.
C. si. (PsV)
T. well (lit. nothing) mh yeah - now if I've understood correctly - your problem is that you sometimes happen to faint (lit. it happens to you to faint).
C. yeah.
After four starters (niente mh si allora) signalling hesitation, se ho capito bene ('if
I've understood correctly') is a 'disclaimer' in Fraser's (1980) sense, a controlling
reformulation or a 'gist' in Thomas' (1989) sense, a 'hedge' in the sense proposed
here, since its scope is the whole illocution. This reformulation is both self-serving
(obeying cautiousness) and altruistic (Fraser, 1980): the client has to check if the
reformulation is correct. It functions as a metacommunicative anticipation-neutralization of a possible disagreement.
Ogni tanto ('sometimes') is a bush, focussed on the propositional content, on the
predication; it reduces the frequency of the symptom thereby also reducing the seriousness of the problem (the fainting). Le capita ('it happens to you') is a bush,
focussed on the propositional content. It is the lexical choice of the predication
which allows the doctor to background the responsibility of the client: the event
'happens', without being brought about intentionally by an agent. On the relational
level, the combination of bushes, globally aimed at minimizing the problem, and
hedges, the hypothetical premise which leaves open the possibility that things can
also be treated differently, constructs the relationship as one where the expert - a
psychiatrist - can also make mistakes. It also weakens the asymmetry and calls for
the client's co-definition of the problem, implicitly stating her control capacity.
5.4. Shields
The scope of mitigation in the following examples is Biihler's (1934) deictic origin, the 'I-here-now' (ego-hic-nunc), at the core of Benveniste's (1970) instance
d'~nonciation. As I have already pointed out under section 4, in shields, there is no
explicit operator of mitigation which works on a more abstract level.
In fact, in shields, the act is not mitigated by explicit linguistic means, but rather
it is dislocated, displaced; there is backgrounding, de-focalization, or even deletion
of the utterance source.
The idea, though it so far lacks a systematic account, has to some extent been
anticipated by the rhetorical notion of aversio (Lausberg, 1967: 431), Greimas and
Courtrs' (1979) notion of d~sembrayage, by Brown and Levinson's notion of
'impersonalization mechanisms' (1987: 273), by Haverkate (1992), and others.
896
As will be clear from the following examples, shields do not work in a scalar way,
but in a yes-no way, releasing their various interactional effects through a process of
contrast with the unmarked, expected, preferred choice. More precisely, there appear
to be some basic strategies of displacement linked with the three basic components
of the deictic origin, the source of the utterance act: the first (based on the 'I'),
which could be labeled 'actantial', to borrow a term from narratology, is represented
prototypically by cases where the act is ascribed to someone else, or by cases where
the author is simply deleted, as in impersonal constructions or agentless passive constructions. Symmetrically, we can have a deletion of the 'you' when no reference is
made to the actual addressee of the message. Objectivization, with impersonal subjects like uno/a, si ('one') in sentences like uno sta in pensiero ('one worries'), is
one of these shields. In discourse production there can be various shifts to and from
the 'I', and there are several ways in which the responsibility for the speech act can
be subtly shifted away, as in Goffman's (1979) cases of 'footing' (Levinson, 1988).
The other two means of displacement are interconnected and can be labeled 'spatiotemporal', since the displacement involves the 'here-now', for instance, when a narrative replaces a discourse focussed on the present context, as in example (13)
below. The displacement can even involve another possible world, opened up
through the use of an hypothetical device, the 'as if', which is very frequent in psychotherapeutic sessions (Gaik, 1992), or a hypothetical sentence: I call these shields
'fictionalization' and 'eventualization' (Haverkate, 1992) shields respectively. These
shields can work locally, at the level of a single utterance, or globally, as textual
strategies.
Example (8) illustrates a non-ego strategy of objectivization:
(8) D. c'~ un'i:perplasia estrogenica - c'~ scritto qui. (SV)
'there's an estrogenic hyperplasia - it is written here'
The scope of the mitigation is the utterance-source, the deictic origin. There is a defocalization of the speaker as the agent of the utterance, which is ascribed to another
impersonal source, that is made more authoritative and unquestionable by the channel (written code) and by the use of the technical register. Shifting the responsibility
to another source - which is not shared by, or immediately accessible to, the
addressee - amounts to weakening the doctor's personal commitment to his diagnosis. Moreover, no reference is made to the two interlocutors: not only is there a 'notI' but also a 'not-you'. There is no reference to the patient either: the disease 'is
there', and it is a fact for which a piece of evidence is invoked. At the same time, the
disease is very clearly identified, with a high-value assignment to the 'precision of
propositional content' parameter. Thus, while the cognitive informativeness of the
act is reinforced, and with it, its argumentative power, the overall relational effect is
one of distancing.
A somewhat similar case, where the doctor shifts to using 'we' (meaning something like 'one' in impersonal constructions), is discussed in Gumperz (1982).
Gumperz claims: "Perhaps .... the underlying motive [for the use of 'we', CC] is to
signal personal distance and to distinguish descriptions of impersonal laboratory pro-
897
898
899
What is mitigated in (12) is the speech act of supposition. This is done in order to
make the (actually realistic) hypothesis of an operation more distant and improbable;
in the background, an evaluation lurks, something like 'maybe an operation will be
necessary'. There is a cumulative use of linguistic devices: textual, i.e. oltreatutto,
which introduces the message as one among other possible topics, thereby decreasing its importance; syntactic, i.e. the subjunctive in the hypothetical construction;
and lexical, i.e. a marker non so and the evaluative adverb paradossalmente. This
latter is a kind of metacommunicative gloss working both on the whole speech act
( ' I ' m only saying, as a paradox, that ... ') and on the propositional content, which is
thus further removed; it also is a stylistic cue that opens the way to a more formal
register.
The shifting away from an aspect of deictic origin can become a global strategy,
as in (13), which serves to introduce the category of topical shields. (13) is an example of a 'narrativization' shield, i.e. a global strategy of de-actualization of a topic,
the patient's present state, which nevertheless is salient in that context, i.e., the first
session in a course of psychotherapy. A negative affect toward the topic can be
inferred from this removal, which signals a separation from the present communication. At another level of analysis, it can be claimed that this strategy aims at self-protection (traditional psychoanalysis would describe it in terms of 'resistance'), and the
client's shield is first of all for herself and her present pain. Whatever the interpretation of the strategy may be, what is crucial is an account of how this actual goal is
achieved, here by shifting a present feeling to the past, which is told in a rather
detached style instead of being enacted. To use Benveniste's famous distinction, the
client's discourse is an histoire which finds it hard to become discours. In narrativization there is an 'I, not-here, not-now' vs. actualization or focalization on 'Ihere-now', which the analyst tries urgently to restore. This type of shield was
focussed by Labov and Fanshel (1977: 336) under the label 'narrative response', as
one of Rhoda's typical mitigation strategies.
(13) P. e poi: dall'85 a11'87 i due anni diciamo in cui: stavo male ma non: per altri
versi ma non cosi male come sto adesso sono comunque riuscita a continuare a lavorare mi hanno cambiato la respo*nsabilit~+
T. *adesso sta male?+in questo momento?
P. be' adesso io sono qua e sono entrata per degli accertamenti: ulteriori=
T. =no dico adesso ora qui.
P. si.
T. sta male?
P. be' - certo.
T. percM parla con me?
P. no: non mi d~ nessun fastidio parlare con lei. (PsS)
P. and then from 85 to 87 the two years when: let's say I felt bad but not in
other respects but not so bad as I feel now anyway I managed to keep working they changed my du*ties+:
D. *now are you feeling bad?+ at the moment?
900
P.
D.
P.
T.
P.
T.
P.
well now I'm here and I've come in for further tests=
=no I mean now at this very moment here.
yes.
are you feeling bad?
well, of course.
because you are speaking to me?
no: I don't mind speaking to you.
(13) is suitable to close this section because it is a transitional case between deictic
shields, in particular 'objectivization', and topical shields, as sketched in section 6
below. The main difference between them is the following. While deictic shields, i.e.
shields based on the negation of one of the aspects of the deictic triad, work by an
overall substitution (e.g. one utterance is substituted by another bearing a non-ego,
non-hic, non-nunc feature), quotational and topical shields operate on 'something'
which actually appears on the linguistic surface. While deictic mitigation is in absentia - deictic shields are in paradigmatic opposition to other unmarked choices in a
given context - quotational and topical mitigation, similar in that to bushes and
hedges, is in praesentia. Let us consider what this 'something' may be.
901
tion, namely, the condition concerning the complete execution of the procedure,
which is related to the scalar parameter 'precision' which affects the propositional
content.
Topical shields are those cases in which there is a strategic backgrounding of a
topic whose occurrence is expected, typically an embarrassing, painful, thorny topic,
through the decreased value assigned to an interactional parameter that can be
labeled 'relevance of a topic for the present purposes of the exchange'. Topical
shields can be further subdivided into strategic digressions, and strategic examples.
Strategic digressions correspond to the rhetoric notion of aversio a materia (Lausberg, 1967: 434), where the mechanism is the 'lateralization' of a topic, obtained
through the use of connectives such as tra l'altro ('by the way', 'what's more'), or
p e r caso ('incidentally', 'by any chance') as in (14):
(14) T. ha avuto per caso qualche altra gravidanza che si ~ interrotta spontaneamente cosl o no ? =
C. =no. (PsV)
T. have you accidentally/incidentally had any other pregnancies with a miscarriage or not?=
C. =no.
The funny effect of (14) for an Italian speaker is due to the ambiguity of the scope
of the Italian mitigator p e r caso (similar to the German zufiillig/zufiilligerweise),
another passe-partout downgrader similar to un attimo. Each of the two possible
English translations selects only one of the two simultaneous possibilities in Italian:
one is "accidentally" ('by any chance'), whose scope is the propositional content;
the other possible translation is "incidentally" ('by the way'), whose scope is the
whole question, presented as a side topic in a hypothetical hierarchy of topics. In this
latter case, it can be classified as a topical shield in which the inferable mitigating
effect is obtained through the 'lateralization' of the topic.
Obviously the topic itself is delicate, and negatively face-threatening for the
patient. Having recognized this, however, we also have to explain how the speaker
manages both to ask the embarrassing question and to protect the addressee's face:
these conflicting goals are simultaneously pursued by the use of the co-textual
shield, which introduces the topic and at the same time downgrades its relevance and
urgency in the context.
Strategic examples are those cases where again the parameter which is reduced is
the 'relevance of the topic for the present purposes of the exchange'. Crucially, this
reduction amounts to a contrast with the expected, preferred choice in that context.
The mechanism by which this reduction is achieved is the paradigmatization of the
topic, obtained through the use of connectives such as ad esempio, p e r esempio ('for
example', 'for instance'), as in (15) (on examples, Caffi and HSlker, 1995):
(15) C. va be' problemi in casa: li ho sempre avuti quindi:
T. che problemi ci sono?
902
C. va be'. ad esempio c'~ mio paph che: *ogni tanto beve +. (PsS)
C. well as to problems at home: I ' v e always had them
T. what are the problems?
C. well for instance there is: my father who *sometimes drinks +.
In other words, mitigation is obtained by downgrading the 'topical relevance' dimension, which, in turn, results from the fact that a topic is placed in a paradigm of possibilities, not presented in a hierarchy. The relational effect at work here is that the
downgrading of the relevance of a topic means implicitly the downgrading of the
emotive salience of the topic for the speaker in the actual communicative situation.
If the topic is actually recognizable as salient in that context, as in (15) the topic of
the father's alcoholism (also mitigated by the bush ogni tanto 'sometimes'), its backgrounding triggers an implicature concerning the unwillingness on the part of the
speaker to handle it. There is a clear similarity between those types of shields and the
rhetorical figure of reticentia.
To sum up: topical shields are the surface means by which the interactional parameter of topical relevance, which is connected by implicature with the emotive
saliency of a topic, is de-intensified. In other words, from the fact that a topic is
given a reduced textual salience - since it is 'lateralized', or set in a paradigm of
equipollent possibilities - it is possible to infer that it is of a minor emotive salience
to the speaker in the given context.
903
Of course, Haley does not mean 'deny' as a logical negation; rather, he regards it as
a qualification of the message which is somewhat incongruent with the message
itself. " B y qualifying his messages with implications that he isn't responsible for his
behavior, a person can avoid defining his relationship with another" (Haley, 1959:
325). By way of opening up a connection between the notions of disqualification and
mitigation in its broad sense, I would like to stress the following points.
Haley breaks down a message into four components, each of which can be
negated: 'I', 'am saying something', 'to you', 'in this situation'. In Haley's exemplification, the cases of 'not-I' are those in which the speaker speaks on behalf of
someone else, for instance, in the name of some authority, of his/her role or status
position, or under the influence of something external (a disease, a drug, etc.). The
speaker, in sum, "may indicate that he is only an instrument transmitting the message" (Haley, 1959: 325), as in example (8) above.
The negation of the 'am saying something' component is exemplified by cases in
which the speaker says something in a contradictory, ambiguous, or unclear way, as
in (12), or signals s/he is not using words but only mentioning them, thereby indicating "one is not communicating a message but merely listing letters of words"
(Haley, 1959: 326), a case close to 'quotational shields'.
The negation of the 'to you' component is exemplified by cases in which the
speaker signals s/he is not talking to the interlocutor but to someone else, for
instance, indicating that s/he is speaking "to the person's status position rather than
to him personally" (Haley, 1959: 326).
The negation of the 'in this situation' component is exemplified by cases in which
the speaker indicates his/her utterance is referring to some other time or place, different from that of the actual interaction s/he is engaged, as in (13). 6
In sum, Haley considers de-responsibilization, i.e. avoiding defining the relationship, as the basic core of incongruent communication, which has its systematic and
dramatic instantiation in schizophrenic behavior. Now, what emerges from the previous sections is that de-responsibilization is also at the core of mitigation. Again, as
I have repeatedly stressed, there is no guaranteed inferential automatism: relational
aspects, like instrumental ones, have to be negotiated between communicators, and
ambivalence is pervasive. However, as I have tried to sketch in this paper, it is possible to specify the ways in which this ambivalence is achieved and reconstruct some
steps through which mitigation can contribute to defining (or not defining) the relationship and the speakers' co-identity. I would suggest that some prototypical cases
of the mitigated choices I have called 'shields', i.e. the displacement of the act to
another source, correspond to Haley's disqualification of the 'I', 'to you' and 'in this
6 Hereis the point of main difference,not made explicit as it should have been, between Haley's model
and Beavin-Bavelas' (1985, 1990) reformulation of it. In fact, 'in this situation' is reinterpreted by
Beavin-Bavelas as 'in this sequence, in this co-text': it is the dimension of evaluation of the message
centered on its sequential relevance, as an (in)adequate, inappropriateresponse to the preceding interaction. This change is probably due to the different kinds of data on which Haley and Beavin-Bavelas
focus their research, i.e. schizophreniccommunication(where the negation of the 'in this situation' component, taken literally, is frequent) and everyday communication.Co-textual relevance is also the main
focus in Sluzky et al.'s (1967) concept of 'transactional disqualification'.
904
situation'. While using different mechanisms, they all seem to entail the benefit of
avoiding the direct assumption of responsibility for the utterance as well as, in
Haley's terms, the cost of not defining the relationship, thereby indirectly defining
the relationship as one where the speaker is not in control. Further, on the relational
level, it is precisely this avoidance at the core of the shield which is potentially contrary to empathy. The extent to which this correlation between shields and emotive
distance can be generalized clearly requires further inquiry. Also needed, are further
constraints which would account for cases like (10) and (11), where shields are, in
fact, empathetic figures of communion. Nevertheless, it can be argued that when
shields increase the emotive distance, as they do in the majority of cases I have studied, this is due to the margin of uncertainty about the definition of the relationship:
Who is speaking to whom? Whose 'project' is enacted by the utterance? On whose
behalf is one speaking? And why does s/he need to hide his/her voice behind another's voice?
Other aspects of Haley's T component, i.e. the speaker as the utterer of an illocutionary act, and Haley's 'am saying something' component, i.e. the proposition
expressed, are simultaneously at play in bushes and hedges. As regards these kinds
of mitigation, the margin of uncertainty about the definition of the relationship is
reduced because there is no displacement to another utterance source or to another
space and time, but only a reduced subscription to the proposition or a reduced
endorsement of the illocution. In other words, there is no disqualification, as in
shields, but a weaker qualification, a weaker claim to truth or a weaker claim to fulfilment (Hiibler, 1983). This weakening may, however, also imply some costs,
although they are not as high as in the case of shields, on the relational level. For
instance, by saying 'probably', with respect to shields such as impersonal constructions, I don't disqualify my message by avoiding responsibility for its content.
Rather, I assume responsibility for reducing the responsibility, even if this means
'leaving the field', at least in the sense that the path is still open to a possible subsequent retraction. On the other hand, a downgrading of the subscription to the proposition, a sort of cognitive withdrawal, may be paralleled by a downgrading of the
relational emotive subscription, a sort of emotive withdrawal. The hearer's co-operation is necessary in order to reconstruct the act on both the content and the relational levels, that is, on both the propositional, cognitive level and on the non-propositional, emotive level. The hearer's problem will also be to decide at what level to
interpret the downgrading: with respect to the content or with respect to the relationship with the speaker? In other words, is the doctor's uncertainty real or strategic, is it related to instrumental aspects or to relational aspects, is it centered, in
Haley's terms, on the 'am saying something' component or rather on the '1', 'to
you', 'in this situation' components? In sum, is the doctor cautious because s/he
really doesn't know or because s/he actually knows but doesn't want to say
(Bergmann, 1992)? It is this twofold uncertainty that is at the basis of the potential
anti-empathetic effects of the use of bushes and hedges in therapeutic contexts. In
conclusion, these mitigation mechanisms may amount to micro-cases of double-bind
since the doctor risks undoing at one level what s/he is doing at another. And this is
particularly true for activity-types such as doctor-patient interaction and psychother-
905
8. Conclusions
Pragmatics offers the theoretical space where psychological, sociological, and
(micro)-linguistic dimensions can be integrated. Compared to the omnipotent subject
of early pragmatics, mitigation projects a wary, tiptoeing, self-effacing subject, on
his/her guard. I think there's no contradiction between these two images, which I see
as complementary. It's precisely because saying is doing that we have to be cautious
and weigh words against contexts.
With regard to my first hypothesis, the examples cursorily discussed above show
clearly that mitigation works at many levels and on many dimensions, affecting
responsibility management in discourse both in cognitive and in emotive terms. The
inferable effects of mitigating devices include both instrumental and relational
aspects, which can be congruent and mutually reinforcing, or somehow in conflict.
With regard to my second hypothesis concerning the gathering of mitigating
means around different scopes in the utterance, it proved useful to enlighten different semantic and pragmatic aspects which can be modified by mitigation, as well as
to clarify the connection between types of mitigation and their relational impact. The
latter was also the subject of my third working hypothesis, which finds external support in theories such as Haley's disqualification, and which however needs to be
tested on more data.
In conclusion, mitigators mitigate because they manage the responsibility of the
speech act in different ways: in the case of bushes, what is weakened is the subscription to the proposition; in the case of hedges, what is weakened is the endorse-
906
ment of the illocution; in the case of shields, what is avoided is the self-ascription to
the utterance, which is then ascribed to another source or shifted to another situation.
While bushes and hedges are scalar devices, e.g., they work along a scale of degrees
of epistemic commitment to the proposition (this is the case with many bushes) or of
degrees of endorsement of one of the scalar dimensions of illocution (this is the case
with hedges), shields are yes-no devices (e.g. ' I ' / ' n o n - I ' , 'now/not-now', etc.) centered directly on the core of the utterance, the deictic origin, the formal support of
subjectivity (l'appareil formel de l'~nonciation, in Benveniste's 1970 terms). Further, while bushes and hedges work in praesentia, i.e. they are lexicalized expressions, shields operate in absentia, by substitution: what is involved is not a downgrading of the quality of some interactional scalar dimensions, but rather, at a more
abstract level, a clash between the co- and contextually bounded expectations and
the actual choice which gives rise to an emotive contrast. The rhetorical categories
closest to bushes and hedges are euphemism, litotes, understatement, and periphrasis, while the closest psychological categories are immediacy (Wiener and Mehrabian, 1968), disqualification (Haley, 1959), equivocation (Beavin-Bavelas et al.,
1990) and, among the emotive devices listed in Caffi and Janney (1994), in particular those of specificity, evidentiality, and volitionality. The rhetorical categories
closest to shields are enallage and reticence, while the closest psychological categories are avoidance strategies (Lewin, 1935), what in psychoanalysis are viewed as
defense mechanisms at work in the 'resistance', and proximity devices (Caffi and
Janney, 1994). Finally, some types of quotational and topical shields can be connected with the rhetorical categories of aversio ab oratore (in particular the case of
sermocinatio, Lausberg, 1967: 432) and of aversio a materia (Lausberg, 1967:
434) respectively. These categories exhibit the strategic mitigating potential of discourse texture, thereby offering new themes for future research on the visibility of
the speaking subject.
References
Adamzik, Kirsten, 1984. Sprachliches Handeln und sozialer Kontakt. Ttibingen: Narr.
Amdt, Horst and Richard W. Janney, 1987. InterGrammar: Toward an integrative model of verbal,
prosodic and kinesic choices in speech. Berlin: De Gruyter.
Asher, R. and J. Simpson, eds., 1994. Encyclopedia of language and linguistics. Oxford: Pergamon.
Austin, John Langshaw, 1962. How to do things with words. London: Oxford University Press.
Authier-Revuz, Jacqueline, 1994. Ces mots qui ne vont pas de soi: Boucles rrflexives et non-coincidences du dire, 2 vols. Paris: Larousse.
Bally, Charles, 1963 [1909]. Trait6 de stylistique franqaise, 2 vols. Gen~ve: Librairie de l'Universit6
Georg & Cie S.A.
Bally, Charles, 1965 [1925]. Le langage et la vie. Genrve: Librerie Droz.
Bange, Pierre, 1992. Analyse conversationnelleet throrie de l'action. Paris: Didier.
Bateson, Gregory, Ray L. Birdwhistell, Ervin Goffman, Edward T. Hall, Don D. Jackson, Albert E.
Scheflen, Stuart J. Sigman, Paul Watzlawick, 1981. La nouvelle communication. Paris: ,~litions du
Seuil.
Bazzanella, Carla, Claudia Caffi and Marina Sbis~, 1991. Scalar dimensions of iUocutionaryforce. In: I.
7.agar, ed., Speech acts: Fiction or reality?, 63-76. Ljubljana: IPrA.
Beavin-Bavelas,Janet, 1985. A situational theory of disqualification: Using language to 'leave the field'.
In: J.P. Forgas, ed., Language and social situations, 189-211. New-York: Springer.
907
Beavin-Bavelas, Janet, Alex Black, Nicole Chovil and Jennifer Mullett, 1990. Equivocal communication. London: Sage.
Benveniste, Emile, 1958. De la subjectivit6 dans le langage. Journal de Psychoiogie. Reprinted in 1966
in: Probl~mes de linguistique g6n6rale. Paris: Gallimard.
Benveniste, Emile, 1970. L'appareil formel de 1'6nonciation. In: T. Todorov, ed., L'6nonciation. Langages 17. Reprinted in 1974 in: Probl~mes de linguistique g6n6rale II. Paris: Gallimard.
Bergmann, Jt~rg R., 1992. Veiled morality: Notes on discretion in psychiatry. In: P. Drew and J. Heritage, eds., 137-162.
Bertalanffy, Ludwig yon, 1968. General system theory. New York: Braziller.
Blum-Kulka, Shoshana, Juliane House and Gabriele Kasper, eds., 1989. Cross-cultural pragmatics:
Requests and apologies. Norwood, NJ: Ablex.
Blum-Kulka, Shoshana, 1992. The metapragmatics of politeness in Israeli society. In: R.J. Watts, S. Ide
and K. Ehlich, eds., Politeness in language. 255-280. Berlin: De Gruyter.
Brown, Roger and Albert Gilman, 1989. Politeness theory and Shakespeare's four major tragedies. Language in society 18: 159-212.
Brown, Penelope and Stephen Levinson, 1978. Universals in language usage: Politeness phenomena. In:
E. Goody, ed., Questions and politeness: Strategies in social interaction, 56-311. Cambridge: Cambridge University Press. Reprinted in 1987 as: Politeness: Some universals in language usage. Cambridge: Cambridge University Press.
Btthler, Karl, 1934. Sprachtheorie. Jena: Fisher.
Caffi, Claudia, 1991. Modulazione, mitigazione, litote. In: M.-E. Conte, A. Giacalone Ramat, P. Ramat,
eds., Dimensioni della linguistica, 169-199. Milano: Angeli.
Caffi, Claudia, 1992. I1 concetto di coinvolgimento nella linguistica pragmatica. In: G. Gobber, ed., La
linguistica pragmatica: Atti del XXIV Congresso della Socie~ di Linguistica Italiana, 267-297.
Roma: Bulzoni.
Caffi, Claudia, 1993. Metapragmatics. In: R. Asher and J. Simpson, eds., The encyclopedia of language
and linguistics. Oxford: Pergamon.
Caffi, Claudia, 1994. Antifrasi, atto illocutorio, atto linguistico, atto locutorio, atto perlocutorio, digressione, enfasi, forza illocutoria, ironia, litote, mitigazione, perifrasi, preterizione, reticenza, riassunto.
In: G.L. Beccaria, ed., Dizionario di linguistica. Torino: Einaudi.
Caffi, Claudia, 1995. Per una prospettiva pragmatica sull'italiano parlato: La mitigazione. Talk held at
Zurich University (unpublished).
Caffi, Claudia, 1997. Mitigation in doctor-patient interaction. In: A. Guerci, ed., Proceedings of the
Third European Colloquium of Etnopharmacology, and the First International Conference of Anthropology and History of Health and Deseases, Genoa, 29 may - 2 june 1996. Genova: Erga Edizioni.
Caffi, Claudia, in preparation. La mitigazione.
Caffi, Claudia and Richard W. Janney, 1994. Toward a pragmatics of emotive communication. Journal
of Pragmatics 22: 325-373.
Caffi, Claudia and Klaus H61ker, eds., 1995. Examples, special issue of Versus, 70/71. Milano: Bompiani.
Cicourel, Aaron, 1992. The interpenetration of communicative contexts: Examples from medical
encounters. In: A. Duranti and C. Goodwin, eds., Rethinking context, 291-310. Cambridge: Cambridge University Press.
von Cranach, Mario, Urs Kalbermatten, Katrin Indermiihle and Beat Gugler, 1980. Zielgerichtetes Handeln. Berne-Stuttgart: Huber.
Dressier, Wolfgang and Lavinia Merlini Barbaresi, 1994. Morphopragmatics. Berlin: De Gruyter.
Drew, Paul and John Heritage, 1992. Talk at work. Cambridge: Cambridge University Press.
Duranti, A. and C. Goodwin, eds., 1992. Rethinking context. Cambridge: Cambridge University Press.
Edmondson, Willis, 1981. Spoken discourse: A model for analysis. London: Longman.
Edmondson, Willis and Juliane House, 1981. Let's talk and talk about it: A pedagogic interactional
grammar of English. MUnchen: Urban.
Fairclough, Norman, 1992. Discourse and social change. Cambridge: Polity.
Fraser, Bruce, 1980. Conversational mitigation. Journal of Pragmatics 4: 341-350.
Frijda, Nico H., 1982. The meanings of emotional expression. In: M.R. Key, ed., Nonverbal communication today: Current research, 103-119. Berlin: De Gruyter.
908
Gaik, Frank, 1992. Radio talk-show therapy and the pragmatics of possible worlds. In: A. Duranti and
C. Goodwin, eds., 271-289.
Ghiglione~ Rodolphe, 1986. L'homme communiquant. Paris: Colin.
Giles, Howard and Nikolas Coupland, 1991. Language: Contexts and consequences. Milton Keynes:
Open University Press.
Giles, Howard, Klaus R. Scherer and Donald M. Taylor, 1979. Speech markers in social interaction. In:
K. Scherer and H. Giles, eds., Social markers in speech, 343-381. Cambridge: Cambridge University
Press.
Givrn, Talmy, 1989. Mind, code and context: Essays in pragmatics. Hillsdale, NJ: Erlbaum.
Goffman, Ervin, 1967. Interaction ritual: Essays on the face-to-face behavior. New York: Doubleday.
Goffman, Ervin, 1979. Footing. Semiotica 25: 1-19.
Greimas, Algirdas J. and Joseph Courtrs, 1979. Dictionnaire de srmiotique. Paris: Hachette.
Grice, Paul H., 1975. Logic and conversation. In: P. Cole and J.L. Morgan, eds., Speech Acts, 41-58.
New York: Academic Press.
Gumperz, John J., 1982. Fact and inference in courtroom testimony. In: J.Gumperz, ed., Language and
social identity, 163-195. Cambridge: Cambridge University Press.
Gumperz, John J., 1992. Contextualization and understanding. In: A. Duranti and C. Goodwin, eds.,
229-252.
Hagrge, Claude, 1985. L'homme de paroles. Paris: Fayard.
Haley, Jay, 1959. An interactional description of schizophrenia. Psychiatry 22: 321-332.
Halliday, Michael A.K., 1976. Modality and modulation in English. In: G. Kress, ed., Halliday: System
and function in language, 189-213. London: Oxford University Press.
Haverkate, Henk, 1992. Deictic categories as mitigating devices. Pragmatics 2(4): 505-522.
Held, Gudrun, 1992. Politeness in linguistic research. In: R.J. Watts, S. Ide and K. Ehlich, eds.,
131-153.
Hickey, Leo, ed., 1989. The pragmatics of style. London: Routledge.
Hill, Jane H. and Judith T. Irvine., eds., 1993. Responsibility and evidence in oral discourse. Cambridge:
Cambridge University Press.
Hrlker, Klaus, 1988. Zur Analyse von Markern. Stuttgart: Steiner.
Holly, Werner, 1979. Imagearbeit im Gespr/~chen. T0bingen: Narr.
House, Juliane and Gabriele Kasper,1981. Politeness markers in English and German. In: F, Coulmas,
ed., Conversational routine, 157-185. The Hague: Mouton.
HiJbler, Axel, 1983. Understatements and hedges in English. Amsterdam: Benjamins.
Kerbrat-Orecchioni, Katherine, 1990-1994. Les interactions verbales, 3 vols. Paris: Colin.
Labov, William and David Fanshel, 1977. Therapeutic discourse: Psychotherapy as conversation. New
York: Academic Press.
Lakoff, George, 1973. Hedges: A study in meaning criteria and the logic of fuzzy concepts. Journal of
Philosophical Logics 2: 458-508.
Lakoff, Robin, 1980. How to look as if you aren't doing anything with words. Speech act qualification.
Versus 26/27: 29-47.
Langner, Michael, 1994. Zur kommunikativen Funktion von Abschw~ichungen. Miinster: Nodus.
Lausberg, Heinrich, 1967 [1949]. Elemente der literarischen Rhetorik. Miinchen: Hueber.
Leech, Geoffrey, 1983. Principles of pragmatics. London: Longman.
Levinson, Stephen, 1987. Pragmatics. Cambridge: Cambridge University Press.
Levinson, Stephen, 1988. Putting linguistics on a proper footing: Explorations in Goffman's concepts of
participation. In: P. Drew and A. Wootton, eds., Erving Goffman: Exploring the interaction order,
161-227. Cambridge: Polity.
Levinson, Stephen, 1992. Activity types and language. In: P. Drew and J. Heritage, eds., 66-100.
Lewin, Kurt, 1935. A dynamic theory of personality. New York: McGraw-Hill.
Lyons, John, 1977. Semantics, 2 vols. Cambridge: Cambridge University Press.
Maynard, Douglas W., 1992. On clinicians co-implicating recipients' perspective in the delivery of diagnostic news. In: P. Drew and J. Heritage, eds., 331-358.
Mey, Jacob L., 1985. Whose language: A study in linguistic pragmatics. Amsterdam: Benjamins.
Mey, Jacob L., 1993. Pragmatics: An Introduction. Oxford: Blackwell.
909
Meyer-Hermann, Reinhard and Rtidiger Weingarten, 1982. Zur Interpretation und interaktiven Funktion
von Abschw~ichungen in Therapiegespr~chen. In: K. Detering, J. Schmidt-Radefeldt and W. Sucharowski, eds., Sprache erkennen und verstehen, 242-252. Ttibingen: Niemeyer.
Mitchum, Patricia A., 1989. Verbal and non-verbal communication in a family practice consultation: A
focus on the physician-patient relationship. In: W. von Raffler-Engel, ed., 109-157.
Mortara-Garavelli, Bice, 1990. Retorica e analisi del discorso. Lingua e stile 15(3): 495-507.
Perelman, Charles and Lucie Olbrechts-Tyteca, 1958. Trait6 de l'argumentation: La nouvelle rhrtorique.
Paris: Presses Universitaires de France.
von Raffler-Engel, W., ed., 1989. Doctor-patient interaction. Amsterdam-Philadelphia: Benjamins.
Sbisa, Marina, 1989. Linguaggio, ragione, interazione. Bologna: I1 Mulino.
Searle, John, 1975. A taxonomy of illocutionary acts. In: K. Gunderson, ed., Minnesota Studies in the
Philosophy of Science VII, 344-369. Minneapolis, MN: Minnesota University Press.
Selting, Margret, 1994. Emphatic speech style - with special focus on the prosodic signalling of heightened emotive involvement in conversation. Journal of Pragmatics 22: 375-408.
Silverman, David, 1987. Communication and medical practice. Social relations in the clinic. London:
Sage.
Simmel, Georg, 1983 [1906]. Psychologie der Diskretion, in Schriften zur Soziologie. Frankfurt am
Main: Suhrkamp.
Sluzki, Carlos E., Janet Beavin, Alejandro Tarnopolsky and Eliseo Veron, 1967. Transactional disqualification: Research on the double bind. Archives General Psychiatry 16, April 1967: 494-504.
Thomas, Jenny A., 1989. Discourse control in confrontational interaction. In: L. Hickey, ed., 133-156.
Verschueren, Jef, 1987. Pragmatics as a theory of linguistic adaptation. IPrA Working Document 1.
Watts, R.J., S. Ide and K. Ehlich, eds., 1992. Politeness in language. Berlin-New York: Mouton de
Gruyter.
Watzlawick, Paul, Janet Beavin and Don Jackson, 1967. Pragmatics of human communication. New
York: Norton.
Weingarten, Riidiger, 1990. Reformulierung in der Gespr~ichspsychotherapie. In: K. Ehlich, A. Koerfer,
A. Redder and R. Weingarten, eds., Medizinische und therapeutische Kommunikation, Diskursanalytische Untersuchungen, 228-240. Opladen: Westdeutscher Verlag.
Wiener, Morton and Albert Mehrabian, 1968. Language within language: Immediacy, a channel in verbal communication. New York: Appleton Century-Crofts.
Winnicott, Donald W., 1965. The maturational processes and the facilitating environment. New York:
International Universities Press.
Claudia Caffi got her tenure at the University of Genoa, Italy, where she teaches in the fields of pragmatics and applied linguistics. Since 1981, she has been one of the editors of the Journal of Pragmatics
(review editor for Continental Europe 1983-1993; since 1993, editor). Since her thesis in semiotics on
J.L. Austin discussed in 1976 at the University of Pavia, where she also got a postgraduate degree in
epistemology, her research work has been devoted to pragmatics, a field in which she has widely published. Her main research interests are: pragmatics, rhetoric, psycholinguistics, applied linguistics (institutional interaction, especially doctor-patient and psychotherapeutic dialogues). Her recent publications
in English include: 'Metapragmatics', and 'Pragmatic presupposition' (in: Asher and Simpson, 1994),
with Richard W. Janney, 'Toward a pragmatics of emotive communication' (in: Journal of Pragmatics
22(3-4), which she co-edited with Richard W. Janney) and Examples, a special double issue of Versus,
70-71 (Milano, Bompiani, 1995) which she co-edited with Klaus Htilker. Currently she is working on a
book on mitigation.