Andrew Aberdein
Department of Logic and Metaphysics
University of St. Andrews
©1998, Andrew Aberdein

Charles Stevenson introduced the term 'persuasive definition' to describe a suspect form of moral argument 'which gives a new conceptual meaning to a familiar word without substantially changing its emotive meaning'. However, as Stevenson acknowledges, such a move can be employed legitimately. If persuasive definition is to be a useful notion, we shall need a criterion for identifying specifically illegitimate usage. I criticize a recent proposed criterion from Keith Burgess-Jackson and offer an alternative.


The term 'persuasive definition' (PD) was introduced by Charles Stevenson to describe a form of moral argument 'which gives a new conceptual meaning to a familiar word without substantially changing its emotive meaning'.1 The phenomenon is a familiar one, and readily identified in the works of the political spin doctor and the advertising copy writer. Aldous Huxley trenchantly describes the grossest aspects of PD: "I always love that kind of argument. The contrary of a thing isn't the contrary; oh, dear me, no! It's the thing itself, but as it truly is. Ask any die-hard what conservatism is; he'll tell you that it's true socialism. And the brewers' trade papers: they're full of articles about the beauty of true temperance. Ordinary temperance is just gross refusal to drink; but true temperance, true temperance is something much more refined. True temperance is a bottle of claret with each meal and three double whiskies after dinner."2 Subtler examples can be found in Gandhi's definition of victory as such that the defeated does not hate the victor, or Matthew Arnold's contention that 'poetry is at bottom a criticism of life'.3 A related and converse phenomenon proceeds by altering the emotive meaning of a term without changing its conceptual meaning. Stevenson calls this 'persuasive quasi-definition'.4

It should be stressed at this point that Stevenson is writing from an emotivist position in ethics. Hence he draws a fundamental difference in kind between conceptual (or descriptive) meaning and emotive meaning, that is between fact and value. Although Stevenson's account of PD played an important role in the articulation of his emotivist theory, it is possible to give it a presentation independent of that theory. For all that is necessary to appreciate such argumentation is a recognition that terms of natural language can carry pejorative or laudatory overtones not explicit in their intension. These are aspects of the meaning of a term that are not relevant to the evaluation of the truth value of any sentence in which it occurs. The distinction is that which Frege draws between Sinn and Färbung, or sense and tone.5 This distinction is familiar and well-entrenched; we will thus employ it henceforth in our discussion of PD, which we may regard as any argument which seeks to profit by altering the sense or reference of a term while preserving its tone. Incidentally, this broadens PD beyond moral discourse into any area where tone may influence argument.

Most discussion of PD has presumed its illegitimacy as a form of argument, and we have seen how it can be turned to discreditable ends. It is, however, worth asking whether this is an inevitable feature of its employment. As Stevenson reminds us, 'not all persuasion is that of the mob orator'.6 It is not unreasonable to seek to employ a terminology with a tone as favourable as possible to the cause one wishes to advance. In some cases this may place sufficient tension on the sense of the words used as to border on PD. The adoption of the slogans 'Pro-Choice' and 'Pro-Life' by the two sides of the abortion debate has just such a character: it is unclear whether or not the sense of the terms 'choice' and 'life' has been unreasonably stretched. However, in so far as neither slogan is likely to mislead, it would seem unjust to convict either lobby of anything worse than efficient PR. Furthermore, we can exhibit clear cut examples of PD—such as that attributed above to Gandhi—which appear quite acceptable. However, even if PD may be employed legitimately, its less wholesome applications demonstrate that in principle it at least stands in need of defence. The unqualified acceptance of such arguments must expose us to a risk of being misled.7

A related problematic aspect of PD is that it is much easier to make than to rebut. If there is no satisfactory way of defending oneself against a charge of PD the notion must be a dangerous one, since it could be used to discredit perfectly reasonable argumentation. Conversely, this would also render it ineffectual; for if it can be applied indiscriminately, it is no longer sufficiently well-defined for its ascription to be informative. Thus if PD is to be presented as a useful critique of good argument, we shall need a criterion for identifying its specifically illegitimate usage which is tightly enough drawn to offer a ready defence to the wrongly accused.

Keith Burgess-Jackson has recently sought to introduce just such a criterion.8 This proceeds from the following account of PD. Firstly, following a suggestion of Stevenson's, he contends that PD is commonly—but not intrinsically—linked to the vagueness of the term at issue.9 Secondly, he holds that such vagueness takes the form of a simple tripartite division of the possible extension of the term into clearly applicable cases, clearly inapplicable cases and a penumbra of borderline cases. Thirdly, he suggests that any PD that changes the status of any case from plainly applicable to plainly inapplicable, or vice versa, is illegitimate, at least 'on the assumption that deceptiveness is bad'.10 By extension, the PD of any term which is not vague, and thus lacks a penumbra, must be illegitimate. Fourthly, he argues that precisification within the penumbra is prima facie legitimate PD, because it admits of the possibility of a bona fide theoretical grounding. He is not perfectly clear on the point, but would seem to regard ad hoc or frivolous precisifications as illegitimate PD. Fifthly, he states that, where backed by independent theoretical motivation, precisification within the penumbra is legitimate PD.

A preliminary point concerns Burgess-Jackson's account of vagueness. Stevenson ascribes vagueness to those terms most especially apt for PD, but he was writing at the very beginning of the modern discussion of vagueness, when that term was employed with less precision than is nowadays customary.11 What is shared by the terms which we saw persuasively defined above (socialism, temperance, victory, poetry), in so far as they are at all imprecise, is not vagueness proper, but a species of ambiguity. Simple ambiguity occurs when a term has two clearly distinct meanings, as with homonyms like 'bank'. Here we are concerned with terms that have many subtly different, but clearly distinct meanings; we may say that they are multiplicitous.12 The crucial difference is that, whereas recognizing the indeterminacy of the borderline is a hallmark of the competent use of vague terms such as tall, short, red, yellow, child, adult, a determination of the borderline is a prerequisite for the competent use of multiplicitous terms, such as socialism, poetry, wisdom, beauty. There may be many different theories as to what socialism is—perhaps as many as there are socialists—but to speak meaningfully of it we must specify one (or at least some range of theories which share all the features salient to our discussion). This contrasts with vague terms, where to precisely specify the borderline is to depart so far from natural usage as to risk unintelligibility ("By 'red' I mean emitting or reflecting visible light of a wavelength greater than x Angstroms"). Burgess-Jackson cites Stevenson as asserting that '[e]thical terms are more than ambiguous; they are vague', but Stevenson is here concerned to stress that the indeterminacy of such terms is much greater than the simple ambiguity of homonymous terms.13 In modern terminology he is describing multiplicity.

A notable aspect of the vagueness/multiplicity distinction is that, whereas vague terms typically also exhibit higher-order vagueness (the boundaries of the penumbra will be themselves vague, as will be the boundaries of any further penumbrae we may introduce), multiplicitous terms do not. At least in principle, we can take the intersection of all the sets of sentences that comprise theories of socialism to form the set of sentences common to all theories,14 and the complement of the union of all such sets to form the set of sentences denied by all theories. The domain of application contains just those definitions of socialism which contain only sentences from the former set; the domain of inapplication contains just those definitions which contain any sentences from the latter set. Hence, if (and only if) the frequently observed feature of PD-apt terms is multiplicity rather than vagueness, then it can be adequately described by Burgess-Jackson's tripartite picture. Hence the drawbacks of the first and second stages of his argument cancel each other out; the inexactitude is merely terminological.

The chief problem with Burgess-Jackson's criterion arises from the fourth and fifth stages, which risk a vicious circularity. The criterion admits as legitimate PD any theoretically motivated precisification into the penumbra, even if the cases at issue only lie within the penumbra because of the theory employed by the proponent of the PD at issue. The legitimacy of the persuasive definer's argument will turn on multiplicity which he is himself responsible for introducing. Hence he may muddy the waters, and then profit from their muddiness. Moreover, any theoretically motivated redefinition of a term, however novel and implausible, will offer the new sense as a further ambiguity of the term and thus broaden the penumbra to include itself. Hence any such redefinition will satisfy Burgess-Jackson's criterion, which collapses into the very weak constraint that PD is legitimate if some theory can be found to underpin the redefinition of the term at issue. Only utterly gratuitous PD would be deemed illegitimate.

What might be done to strengthen this criterion? We could attempt to forestall the circularity by placing a constraint on the sort of multiplicity to be regarded as admissible. Hence we could specify that the multiplicity should be antecedent to the PD. But it would be difficult to state what degree of antecedence would be required. The mere priority of the multiplicity to the first case of PD would not be enough, since this would only differ from Burgess-Jackson's criterion in requiring that the theory was specified first. Antecedence of the multiplicity to the entire theory would be very hard to determine, and would seem to take us on an irrelevant excursion into the history of ideas. Alternatively, we could specify that the multiplicity was independent of the PD. But this would suffer from similar problems: we should be obliged to assess the rather speculative counterfactual that the cases at issue would not have been within the penumbra of multiplicity if the theory had not been advanced. There may be some scope for a working criterion here, but there is also an impression that the point of the problem is being missed.

In particular the focus on multiplicity places an undue concentration on semantic features of the term, when what is most important is the role that it plays in argument. We can see this more clearly by focussing on the quotation from Gandhi which we identified above as a possible example of legitimate PD. This example would seem to be classified as illegitimate on any version of the Burgess-Jackson criterion considered so far. 'Victory' is not obviously an ambiguous term (at least in so far as any term can be unambiguous); unless Gandhi's remark makes it so.15 Even if we take this ambiguity to be sufficient for the remark to count as legitimate PD on Burgess-Jackson's original, weak criterion, it is neither antecedent to, nor independent of, Gandhi's usage. Hence neither strengthening respects our intuitions about PD.

Why does this remark seem to be a legitimate employment of PD? What seems crucial is that the redefinition suggests to us (perhaps abetted by our knowledge of Gandhi's reputation) how an argument might be constructed for using the term in the new way. Burgess-Jackson is certainly heading in the right direction with his appeal to independent theoretical motivation. PD is a promissory note for such justification, more specifically such that it warrants the new definition. But it is not enough merely for it to exist,—the definer's interlocutor must appreciate its force, even if he is not disposed to accept the definition. He must recognize the definition for what it is. Otherwise, if the interlocutor is ignorant of or hostile to the definer's theory, the PD can only serve to mislead.

This suggests a new criterion for legitimate PD. PD amounts to an invitation to one's interlocutor to accept without argument (if only for the purposes of discussion) the definitions of one's own theory. This is something he may understandably be reluctant to do, especially if these definitions conflict with ones offered by a theory of his own. The crucial problem afflicting illegitimate PD is a failure to engage with one's interlocutor's understanding of the term at issue. Mere insistence on the use of one's own theory, and the definitions that support it, without explanation of why this is superior, cannot hope to be legitimately persuasive. PD can only be legitimate if it is undertaken in a context in which its theoretical indebtedness is made clear, and the redefinition of the term by that theory justified. The precise character of the required context will vary, depending upon the goals of the persuasive definer. We would expect a commensurately more serious degree of justification for PD employed in serious and earnest argument, than for PD advanced in a more allusive, thought-provoking or aphoristic manner. We may be satisfied with little explicit defence of the redefinition if the PD serves only to redirect our attention to some aspect of the definiendum, perhaps by presenting an apparently paradoxical or surprising definiens. (This latter category is what Stevenson calls re-emphatic definition.16 ) Where the success of subsequent argument turns on our acceptance of the PD we are likely to be more cautious. What we will require from the proposer of such a definition is a reason why the new sense should be attached to the disputed term, rather than to some neologism.

This more general problem, of defending the legitimacy of a change of meaning, is one that is well known in the philosophy of science. As scientific theories develop, their terms are often redefined, in the hope that such conceptual sharpening will yield a clearer description of the subject matter. Hence, a Copernican definition of 'planet' would include 'body that follows an elliptical orbit around the sun', whereas a Ptolemaic definition would include 'body that follows an epicyclic orbit around the earth'. A criticism that such developments must answer is that they are no longer talking about the same thing. The natural response to this is to indicate some fundamental feature which both theories capture. For the Copernicans this would be easy; they could simply point at the planets. For more abstract or complex concepts, such as those employed in the pure or social sciences, an appeal to an underlying ostensive definition is often not readily available, or even possible.

An alternative is to appeal to some core meaning, essential to the correct use of the term, which the redefinition preserves. Precise articulation of such a core meaning may be unattainable, but, if there is sufficient general agreement over it, the legitimacy of the redefinition should be resoluble. Burgess-Jackson's domains of clearly applicable and inapplicable cases might be suggested as a putative core meaning. The problem with this is that it aims to give a purely extensional account of an intensional problem. As such it is not sufficiently explanatory to guard against the risk of being merely ad hoc. Indeed, as we have already seen, a core meaning such as this would be either vacuously preserved or, on a strengthened account, misleadingly not preserved. Prior to, or independent of, Copernican cosmology a satellite of the sun would not count as a planet. What is needed is something that might be regarded as a minimal definition of the disputed term.

Our discussion of PD would benefit from an example with a more sustained context; one such is provided by Burgess-Jackson. The principal concern of his article is with the debate between radical feminists and other parties, most notably liberal feminists, over the definition of rape. The liberal definition is that rape is sexual intercourse in which one party does not offer informed, rational consent. The radical feminists contend that rape can only be understood as serving to reinforce a patriarchal hegemony. Hence they can define rape as existing 'any time sexual intercourse occurs when it has not been initiated by the woman out of her own genuine affect and desire' and be prepared to embrace the consequence that 'under this definition, most of the decently married bedrooms across America are the settings for nightly rape'.17 Critics of the radical feminist position have argued, explicitly and implicitly, that this constitutes illegitimate persuasive definition.18

Burgess-Jackson argues that his account of PD refutes this contention. On his terms this is certainly so. Since the radical feminist definition of 'rape' has been seriously advanced, all the cases which it identifies as rape must lie in the penumbra or positive extension of the term. So, for Burgess-Jackson, the radical feminists cannot be engaged in illegitimate PD. But, as we have seen, this account risks utter vacuity; it makes it sufficient that the radical feminist definition has been advanced for it to fail to be illegitimate PD. The only way it could be illegitimate is if it were without theoretical basis, which is certainly not the case. On a strengthening of Burgess-Jackson's criterion we should require that 'rape' was sufficiently multiplicitous prior to or independent of the radical feminists' definition for all of the cases they identify as rape to be included at least within the penumbra. It is unclear whether this is a challenge that they could meet; the surprise that their definition has elicited from commentators suggests that it is not. But we have seen that the strengthened criterion can be unduly narrow.

What of our criterion? What we would require from the radical feminists is that their justification for the continued use of 'rape' to refer to the phenomenon which receives that name in their theory always be made clear. Here we must distinguish, as Burgess-Jackson does not, between the charge that the radical feminist definition is always illegitimate PD and the charge that it is sometimes so. The latter would seem proven, for it is certainly possible for radical feminists to speak in such a way that their interlocuters are understandably misled as to the extension of the term, and for them to profit by this confusion. For the stronger charge to be established, we should need to show that the radical feminists were unable or unwilling to defend their definition, by the lights of their theory, so that their interlocutors were always misled as to the context in which the definition was advanced. Since we are unlikely to be able to show this, we should conclude that the radical feminist definition of rape can be legitimately advanced, if sufficient care is exercised.

In summary, we have (1) offered a broader account of PD in terms of sense and tone, rather than conceptual and emotive meaning; (2) exhibited the importance of a criterion for distinguishing between legitimate and illegitimate PD; (3) examined, and ultimately rejected such a criterion based on the multiplicity of the term at issue and (4) offered a new criterion turning on the theoretical defence of the novel use of the term, by appeal to the preservation of a core sense.19


1. Stevenson 1938 p. 331.

2. Huxley Eyeless in Gaza, cited in Stevenson 1944 p. 214; Huxley's emphasis.

3. Gandhi, attributed in Pawlowski 1971 p. 474; Arnold Essays in Criticism (London, 1888).

4. Stevenson 1944 p. 278.

5. Dummett 1973 pp. 2f. 'Tone' is the conventional, if imperfectly literal, translation. Frege also uses the term Beleuchtung.

6. Stevenson 1944 p. 215.

7. Some people designate only the illegitimate cases as PD; but it seems an unnecessary complication to deny that the legitimate cases share the same form. Hence we will characterize the distinction as between legitimate and illegitimate PD.

8. Burgess-Jackson 1995 (reprinted in Burgess-Jackson 1996).

9. Ibid. p. 430, citing Stevenson 1938 p. 333.

10. Ibid. p. 433.

11. Williamson 1994 pp. 70ff.

12. This sort of ambiguity is closer to Sainsbury's 'incompleteness' than to homonymy, although he only discusses a much simpler example, where his terminology seems more appropriate (Sainsbury 1987 p. 38).

13. Burgess-Jackson 1995 p. 431 n43, citing Stevenson 1944 p. 34; Stevenson's emphasis.

14. Of course, this may be the empty set, in which case we would have shown socialism to be a family resemblance concept. This is subsumed under the general account of multiplicity.

15. Richard Robinson, for instance, claims that all words are ambiguous (his Definition (Oxford, 1954) cited in Burgess-Jackson 1995 p. 431 n42).

16. Stevenson 1944 p. 290.

17. Robin Morgan 1980 'Theory and Practice: Pornography and Rape' in Laura Lederer, ed., Take Back the Night: Women on Pornography pp. 136f, cited in Burgess-Jackson 1995 pp. 427f.

18. Burgess-Jackson cites several such claims, most notably that in Norman Podhoretz 1992 'Rape in Feminist Eyes' Commentary 93. Rather forbiddingly, he adds that '[n]o philosopher of whom I'm aware, acting as a philosopher or in any other capacity, has claimed, much less argued, that radical feminists are engaged in PD' (Burgess-Jackson 1995 p. 426, my emphasis).

19. I am grateful for the discussion of earlier drafts of this paper with Stephen Read and Patrick Greenough.



Burgess-Jackson, K. "Rape and Persuasive Definition" Canadian Journal of Philosophy 25 (1995) pp 415-454.

_________. Rape: A Philosophical Investigation (Brookfield, VT: Dartmouth, 1996)

Dummett, M. Frege: Philosophy of Language (London: Duckworth, 1973)

Pawlowski, T. "Persuasive Definitions" Scientia 106 (1971) pp 473-481.

Sainsbury, R. M. Paradoxes (Cambridge: Cambridge University Press, 1987; 2/e 1995)

Stevenson, C. L. "Persuasive Definitions" Mind 47 (1938) pp 331-350.

_________. Ethics and Language (New Haven, CT: Yale University Press, 1944.)

Williamson, T. Vagueness (London: Routledge, 1994.)

 View Commentary by B. Russell

View Index of Papers and Commentaries

Return to Main Menu