Вы находитесь на странице: 1из 7

302 INTERNATIONAL RELATIONS 20(3)

On Legitimacy
Friedrich Kratochwil, European University Institute, Florence

Abstract
This article provides a conceptual analysis of different uses of the term ‘legitimacy’.
Rather than attempting to provide a simple ‘definition’ I argue that the meaning of the
concept cannot be understood in terms of a clear reference, but has to be analysed through
its links with other concepts within a semantic field. In this way we are also able to
‘explain’ why at first contradictory uses of the term (designating its ‘input’ or the ‘output’
side) are part of its ‘grammar’. To that extent attempts at stipulative definitions aiming
at an unequivocal sense of the term fundamentally misunderstand the function of this
concept within the political discourse.

Keywords: conceptual analysis, contestable concepts, forms of rule, legitimacy

An inquiry into the meaning of ‘legitimacy’ seems at first rather straightforward. In


accordance with the standard criteria of social scientific analysis we are to determine
the elements in the ‘real world’ which the terms in our conceptual scheme or theory
‘represent’. To that extent we can say that we know what the word ‘legitimacy’
means when we are able to clearly identify and match the indicated elements or
factors in the world ‘out there’. However, even a moment’s reflection suggests that
the ideal of reducing a conceptual analysis to some form of matching operation
runs quickly into heavy waters. There is first the problem that we use the term
for decisions which are the result of certain commonly accepted procedures, such
as voting, or that a binding decision has been taken by a duly authorized agent.
However, we also consider decisions legitimate when we approve of the specific
result, even if the latter has a rather defective pedigree of procedural correctness.
Especially in complicated ‘multilevel’ negotiations and governance issues familiar
from EU policy-making, the ‘output’ legitimacy of the subsequent regulations is
frequently stressed, even if the ‘democratic deficit’ of the mode of arriving at the
decision is admitted.1
Although input and output legitimacy seem to point in different directions, one
still could argue that the ‘acceptance’ by the audience provides the core meaning
of legitimacy. But on closer inspection we have to realize that a second problem
suddenly arises, or that the original problem simply has moved and reappears now
in a different guise. While input legitimacy identifies clearly ex ante the factors that
are bestowing validity on a decision (and thereby a presumably sufficient ‘pull’ for
accepting it), it seems difficult, not to say impossible, to designate ex ante the fac-
tors that make a specific outcome ‘legitimate’. This seems to be the case because a
variety of procedural and substantive criteria interact in apparently unforeseeable

International Relations Copyright © 2006 SAGE Publications


London, Thousand Oaks, CA and New Delhi, Vol 20(3):302–308
[DOI: 10.1177/0047117806066706]
Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015
ON LEGITIMACY 303

ways (quite aside from the fact that common exhaustion and unwillingness to reopen
issues which are probably the most frequent elements of ‘final’ deals).
Particularly in those latter cases the representational ideal of input legitimacy
can only partially be met, if at all; and output legitimacy seems simply at odds with
it. Thus our suspicions of a conceptual quagmire appear to be borne out, especially
when we remember cases in which procedural correctness is not able to defeat more
substantive concerns, as when we argue that a purely ‘legal’ argumentation does not
take care of the more substantive concerns that are being put forward in order to sever
or at least undermine the presumably tight link between legality and legitimacy.
After all, a law which is passed in accordance with all the formal requirements
but which entails the deprivation of some basic rights for a minority will hardly be
considered legitimate, even if many do not want to doubt its legality.
Finally, there seems to exist some form of ‘legitimacy’ which works largely via
extension, i.e. by linking or analogizing a particular decision or alternative to an
already widely accepted legitimate normative interpretation or practice. Thomas
Franck has suggested that in international law in particular, where effective law-
making and central decision-making institutions are lacking or are only weakly
articulated, ‘coherence’ serves as an effective legitimizing force.2 But again, such
extensions which use interpretations and analogies governed by the rules via arte
legis meeting this procedural standard is at best a necessary condition for the validity
of the new norm, rather than a fully fledged acceptance. In the discussions we hear
then often the argument that such extensions arte legis are only inchoate claims (de
lege ferenda) as opposed to well-established norms.
The above considerations make it also difficult to assert – in accordance with both
the descriptive ideals of conceptual analysis and the basic notion of contractualism
– that ‘assent’ represents the common core of all varieties of legitimacy. Since
assent is a formal act framed by certain procedural requirements – at one extreme,
even silence might be construed as assent while, at the other, explicit and formal-
ized actions such as the ‘signing’ of a document have to be undertaken – the essen-
tially procedural nature of legitimacy could then be established. While there is
something to the argument as it recognizes implicitly the importance of speech acts,
such an interpretation nevertheless mistakes their nature by pressing them into an
inappropriate framework and committing therefore a ‘descriptive fallacy’.3 Some
further clarifications are, therefore, necessary.
After all, if we adduce arguments of legitimacy we are not simply stating our pre-
ferences but suggesting that our ‘acceptance’ of a certain proposal or alternative can
be expected to be adhered to by others because of its reasonableness (all things con-
sidered), by its minimal disturbance of ongoing practices or the necessary sacrifices
of other values, or by its expected overall benefit to the community. Such an assent
cannot be based on assent on purely idiosyncratic grounds, since here we are not deal-
ing with the coincidence of personal preferences, but – as is common in politics –
with the establishment of a rule or policy that is binding upon all. Even if thereby
‘winners’ and ‘losers’ are created, we expect compliance from both equally. To
that extent only claims that satisfy at least potentially the generalization criterion

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015


304 INTERNATIONAL RELATIONS 20(3)

(everyone in that situation ought to accept x for reasons a through f ) are usually
admissible, as other self-serving arguments would only add insult to injury. Here
Rousseau’s distinction between the volonté de tous and the volonté générale is
relevant, as are his attempts to specify the necessary criteria.4
An adherent of the proceduralist view of legitimacy could now argue that, on
second thoughts, many of the more substantive reasons making for output legitim-
acy turn out to be results of the further specifications of the conditions which make
the set of alternatives legitimate in the first place. Thus, contract law specifies, for
example, that the rights of third parties must not be affected for the contract to be
valid, making it appear that the meeting of the wills establishes the validity of the
contract only when done in accordance with certain procedural requirements. As
we know from contract law, such a validity is only presumptive and many other
factors might influence the actual validity such as consideration about the ‘fair’
allocation of risk, policy considerations, issues of whether or not the contract falls
under the contra bonos mores exception, etc. Needless to say, in political debates
such ‘technical’ arguments concerning the specifications of ends will play a less
prominent role, except when constitutionality issues and previous relevant ‘cases’
are cited.
Before we rush to a conclusion, it is not true that all agreements need to satisfy
some form of universalizability or intersubjectively shared reasons in order to be
legitimate. We should remember the following: an agreement on a norm such as
in a pure game of coordination – such as the rule dividing a beach into exclusive
zones for ball-playing, sunbathing or picnicking – does not seem to require such
a test. Nobody cares for what reasons the rule was finally adopted or where the
borders’ lines – subject to conditions of non-crowding – are drawn on the beach,
as long as s/he is able to enjoy his or her preferred activity. Here idiosyncratic
reasons and purely personal preferences seem to be sufficient reasons for adoption
and ‘self-enforcement’ of the rules. To that extent, issues of legitimacy hardly arise.
Similarly, when the presumption of ‘acceptance’ is a matter of natural or logical
necessity, such as in a mathematical problem, where we do have to follow certain
procedures, it would be a circumlocution or downright odd to argue that we have to
accept the solution because it was arrived at by a certain prescribed procedure. The
appropriate vocabulary is therefore one based on the distinction of ‘correct’ and
‘incorrect’ rather than on ‘legitimate’ and ‘illegitimate’.
The last two cases draw attention to the fact that issues of legitimacy are clearly
voiced only when practical matters are at issue and when, in the absence of a con-
sensus, no evident and compelling algorithm can be found for arriving at a decision.
The notion of ‘legitimacy’ points, therefore, in its intrinsic links to conceptions of
‘discretion’ and ‘judgement’ on the one hand, and to ‘rule’ (in the sense of making
decisions that are binding on all) and power (in the sense of compellence) on the
other hand. The concept serves thus as a conversation stopper, by providing reasons
for the presumptions of why a decision or a law ought to be supported, but it is also
a ‘conversation opener’, when, for example, the legitimacy based on procedural
grounds is subjected to scrutiny and the implied presumption is weakened on factual

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015


ON LEGITIMACY 305

or normative grounds. The law might be a ‘bad law’, as it does not bring about the
desired effect, it might infringe on other values or liberties of the constituents, or
the set of choices from which the decision was made lacked integrity. Finally, the
discretion granted to a public official might have been abused, since the decision
was not based on the relevant evidence, or might have relied on questionable stand-
ards for assessing or distributing the risks involved. Thus, in the light of these more
substantive considerations the presumption of the validity of the decision is over-
turned despite its prima facie claims to legitimacy.
From these conceptual claims about legitimacy, several corollaries follow. One,
legitimacy is badly understood when it is treated as a descriptive term rather than
one of appraisal. While it may seem that ‘legitimate’ functions like a term that is
supposed to ascertain a ‘property’ of a decision or law – perhaps analogously to
colour terms in the case of objects – our discussion above showed why such an
inference is mistaken. Reducing the operations involved in appraising to one of de-
scribing is likely to engender the same difficulties that we encounter in the ascrip-
tion ‘goodness’ if we treat the latter as a simple ‘property’ of an object or action.
Whatever we might think of these arguments, a second corollary of the above
considerations is that the meaning of appraising terms does not follow the logic of
a simple ‘matching’ operation familiar from the traditional debates about concept
formation and their ‘operationalization’.5 Rather it is better captured by the notion
of ‘use’ à la Wittgenstein.6 The meaning of the term consists in its use according
to recognized and shared criteria not in its representational capacities of the world
‘out there’. Thus if someone says ‘This is a good car’, thereby communicating his
commendatory judgment, he always opens himself to the further queries of ‘Why
do you say so?’. When he answers ‘Because it is yellow’, we know that he is either
pulling our leg or is acting like an ‘idiot’, i.e. has a private language and refuses
to participate in common communications. It does not matter at all if in the latter
case he is – contrary to Hume’s argument concerning the function of these claims –
honestly describing his actual likes and reveals thereby his preferences.
A third corollary for conceptual analysis results from the problematic usage of the
term ‘legitimacy’ for both procedures (input legitimacy) and actual results (output
legitimacy). This seems to violate classical standards for taxonomies that require
exhaustive categories and attributions according to a common characteristic that all
cases brought under a description are supposed to share. But given that legitimacy
like ‘power’ or ‘authority’ seems to represent a cluster of different phenomena,
there might not be a common core or one that is trivial such as when we say that
we use ‘power’ when we want to show that someone or something had an effect
on something else. Wittgenstein’s argument about a rope that consists of different
intertwined threads, one upon another with no one string7 going through the whole
rope, results in quite a different set of taxonomic criteria, which are far from
arbitrary even if clearly at odds with classical Aristotelian logic. As the ‘cognitive
revolution’ in social psychology8 has shown, instead of focusing on one central
common property, we usually classify according to a variety of characteristics of
which some ‘cases’ share a few but not necessarily all features of a paradigmatic

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015


306 INTERNATIONAL RELATIONS 20(3)

example.9 Concepts are derived from paradigmatic best cases and applied to different
contexts by analogy (rather than through the ascertainment of an essential feature).
This makes for the ‘fuzzy boundaries’ of the concepts and for the contestability of
their ascription,10 because in the end no single feature might be common to all cases
instantiating this concept. Consequently, the assignment might resemble more a
grading decision than strict observation or matching operation.
This leads me to the fourth corollary. The task of conceptual analysis is not to
resolve the conceptual ambiguities by definitional fiat (as in the case when data-
banks have to decide whether or not England or the US was ‘democratic’ in the
early nineteenth century when neither had universal suffrage nor meaningful con-
tested elections), but rather to discover the ‘grammar’ of the concept, i.e. show
how it functions in the political discourse. Consider in this context the concept of
‘authority’ that shows some striking similarities to problems we encountered in
our examination of legitimacy. Here too the term is used for two quite different
phenomena.
On the one hand, we speak of an ‘authority’ when we want to stress the compet-
ence and judgement of a person in a certain field. In accordance with the etymology
of the term this version stresses the non-positional and non-institutional sources of
power and influence, as opposed to both technical expertise or formal power. As
opposed to a formal imperium or potestas the ‘authority’ (auctoritas) of, for example,
the princeps in Rome originally meant simply that a decision or suggestion made
by this person had special ‘weight’.11 In other words, the reasons for deciding in
favour of the suggested alternative were ‘enhanced’ (augere = to enhance, increase,
strengthen) because they came from that person. Similarly, when we say that a
person is an authority in a certain field we stress his/her knowledge, the person’s
standing in the field, etc., and imply with this characterization a presumption that
his or her proposal or decision was of a certain quality and that, therefore, it should
be trusted and followed.
On the other hand, we also use speak of authority, as Peter Blau in his critique
of Weber pointed out a long time ago,12 when we mean that a particular person is
in authority, i.e. has certain powers and exercises them as part of his or her holding
an office. Here questions of competence seem to be backgrounded and the ‘insti-
tutional’ fact of power is stressed, analogously to the institutional features in the
case of input legitimacy. The best example is the command by an officer that has to
be obeyed even if it is clearly not effective for attaining the goal of a mission.
While these two notions of authority are rather different, and seem to emphasize
at first contradictory elements, they are not contradictory at all if one understands
the ‘grammar’ of the term. It lets us see why and how such seemingly disparate
notions were lumped together. Here Weber himself provides an important clue.
Although he stresses the institutional power inherent in rational legal forms of
authority, he also reminds us that in the case of rational legal rule, holding an office
is dependent upon a certain formal training and distinct career. This training is not
only to impart a special knowledge and competence to the future bureaucrat, as, for
example, ascertained by competitive examinations as a precondition for entering

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015


ON LEGITIMACY 307

the civil service, it is also designed to play down, or even eliminate, personal or
familial factors that were of such decisive importance in traditional forms of rule.
While, of course, the reduction of practical judgement to some form of ad-
ministrative or technical competence is problematic – ‘functionalist’ arguments
notwithstanding – we can see now more clearly why and how the at first so dissimilar
cases are conceptually and historically related. Being in authority is justified by
somehow having become an authority. Finally, as Weber’s own discussion of
legitimacy shows, the meaning of the term is not approached via definitions and
classical taxonomic procedures, but by placing the term in the context of other
concepts. Thus legitimacy is discussed in conjunction with ‘rule’ or dominium and
with ‘authority’, social conduct, belief, convention, law, etc.13 To that extent Weber
also suggests that understanding a concept does not proceed according to some
form of simple matching operation between ‘reality’ and our terms of discourse, but
by systematically showing the links of the concepts to other concepts placing it in
a semantic field. This has important implications for the study of rules. As we have
seen, rules establish meanings not only by providing the criteria of the use of a term,
they also allow us through their conceptual links ‘to go on’14 by bridging different
activities and practices. As Searle suggested, we do not understand the concept of
money, for example, if we reduce it to its most obvious function (exchange), or
to the palpable properties the tokens serving as money may possess. Rather we
know what the concept means when we are able to see its connection with a host of
practical activities: saving, purchasing, lending, investing, servicing debts, etc.15
One could now argue that ‘money’ is not an appraisal term. Nevertheless, I main-
tain that the above remarks about linkages and semantic fields are true of social
(institutional) concepts in general, only that in the case of appraisals additional
problems in regard to field-dependent criteria arise, as discussed above. In this way
we also understand now – and are no longer surprised – why there exists a tendency
to reduce concepts of appraisal to simple institutional terms, as we saw in the case
of (input) legitimacy.

Notes

1 On the distinction between input and output legitimacy, see Fritz Scharpf, Regieren in Europa
(Frankfurt and New York: Campus, 1999); Arthur Benz, ‘Politische Steuerung in lose gekoppelten
Mehrebenensystemen’, in Raymund Werle and Uwe Schiman (eds), Gesellschaftliche Komplexitaet
und kollektive Handlungsfaehigkeit (Frankfurt: Campus, 2000), pp. 97–124.
2 See Thomas Franck, The Power of Legitimacy among Nations (New York: Oxford University
Press, 1990).
3 For a good discussion of speech acts and institutional rules that work together and thus function
quite differently from simple descriptions, see Wouter G. Werner and Jaap de Wilde’s discussion of
the notion of ‘sovereignty’ and its role in the political discourse in ‘The Endurance of Sovereignty’,
European Journal of International Relations, 7, September 2001, pp. 263–314.
4 Jean Jacques Rousseau, Du contrat sociale, trans. and ed. Chester Crocker (New York: Washington
Square Press, 1971), Book II, ch. 3.
5 For a critical view of the usual methods of operationalizing terms which stays, however, within the
classical canon of logical conceptual analysis, see Giovanni Sartori, ‘Concept Misinformation in
Comparative Politics’, American Political Science Review, 64(4), 1970, pp. 1033–53.

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015


308 INTERNATIONAL RELATIONS 20(3)

6 For an extensive discussion of these points in regard to political concepts, see Hanna Pitkin,
Wittgenstein on Justice (Berkeley: University of California Press, 1972).
7 See Ludwig Wittgenstein, Philosophical Investigation, trans. G. E. M. Anscombe (New York:
Macmillan, 1953).
8 See George Lakhoff, Women, Fire and Dangerous Things (Chicago: University of Chicago Press,
1997).
9 For an important discussion of these problems, see James Davis, Terms of Inquiry (Baltimore, MD:
Johns Hopkins University Press, 2005).
10 On the issue of the contestability of political concepts, see William Connolly, The Terms of
Political Discourse, 2nd edn (Princeton, NJ: Princeton University Press, 1983).
11 See Theodor Eschenburg, Ueber Autoritaet (Frankfurt: Suhrkamp, 1965).
12 Peter Blau, ‘Critical Remarks on Weber’s Theory of Authority’, American Political Science
Review, 57 (June 1963).
13 See Max Weber, Wirtschaft und Gesellschaft, ed. Johannes Winckelmann, 2 vols (Cologne:
Kiepenheuer und Witsch, 1964), vol. 1, pp. 1–42, Soziologische Kategorienlehre.
14 Wittgenstein, Philosophical Investigations, para. 208.
15 See John R. Searle, ‘What is an Institution?’, Journal of Institutional Economics, 1( l ), June 2005,
pp. 1–22.

Downloaded from ire.sagepub.com at NANYANG TECH UNIV LIBRARY on June 5, 2015

Вам также может понравиться