Minimalism, Fiction and Ethical Truth

interesting or important difference between the claim that p and the claim that 8p 9 is true; and (ii) that the disquotational property of truth predicates, manifested in the truth-schema 8p 9 is true iff p, is the only theoretically interesting or important primitive property of those predicates. However, it is dubious that this 8minimalist core 9 characterises a single position which forms a useful subject for philosophical investigation; rather, further commitments must be added in order to provide 2 precise targets for scrutiny. And what goes for truth goes for other semantic notions as well: 8minimalism 9 covers such a multitude of sins that nothing but confusion is generated by generalisations which fail to attend to the particularities of each sin.<br><br> In this paper, I propose to investigate that kind of minimalist view which is elaborated and defended by Crispin Wright in his book Truth and Objectivity and subsequent publications. I shall be particularly concerned to argue that no satisfactory position emerges from those writings; however, I shall also suggest that it is quite unclear what patches could be applied 3 i.e. I shall suggest that Wright 9s minimalism belongs to a family of views none of which is satisfactory.<br><br> Each of the members of this minimalist family is committed to the following four, admittedly somewhat vague, theses: (1) Variation in Semantic Width : Semantic predicates vary in width 3 so, e.g., truth and (maybe?) truth-aptness are thicker, or more robust, for some areas of discourse than they are for others; (2) Minimal Truth : Minimal truth does not make demands which go much beyond the demands of warranted assertability -- e.g. in appropriately restricted domains, superassertability 1 coincides with truth; (3) Minimal Truth-Aptness : A sufficient condition for (minimal) truth-aptness for sentences is provided by satisfaction of certain constraints on fit substituends in the truth-schema, viz: (a) they must have the correct syntax -- roughly, having that declarative form which fits them for embedding by negations, conditionals, verbs of 1 Roughly, superassertability is the property of being warranted by some state of information and remaining warranted no matter how that state of information is enlarged or improved upon -- c.f. Wright (1984:330).<br><br> Whether there is a useful notion here -- and exactly how Wright 9s rough formulation should be improved upon to identify that notion -- are questions which I shall not take up. 3 propositional attitude, etc. -- and (b) they must be disciplined by norms of appropriate utterance.<br><br> 2 ; (4) Objectivity Reconceived : Acceptance of (2) and (3) forces a reconceptualisation or refiguration of the nature of local debates between realists and anti-realists, since non-cognitivisms are ruled out, and error theories are extraordinarily difficult to justify; thus, e.g. in the case of ethics, (2) and (3) entail that ethical non- cognitivisms are ruled out, and that Mackie 9s error theory is utterly implausible. One strategy which one might take in arguing against the family of views is to argue against a particular thesis from this list.<br><br> I shall begin with some suggestions about the construction of cases against each thesis (Part 1); however, these suggestions will be primarily in the nature of preliminary skirmishes conducted with a view to clarifying some of the important vocabulary involved. 3 Another strategy which one might take in arguing against the family of views is to argue that there is something unsatisfactory about the package of theses as a whole. This is the form of argument which I wish to pursue, in connection with questions about reference (Part 2).<br><br> In particular, I shall argue that no distinctive and plausible view about singular term reference and predicate reference coheres with, or emerges from, the package of theses. The intuitive thought which motivates the argument is this: If truth and truth-aptness are (relatively) easy to achieve -- as our minimalists suppose -- then so is reference. But (relative) 2 A stronger version of this thesis holds that the stated conditions are also necessary for truth- aptness -- see, e.g., Boghossian (1989).<br><br> Must sentences have the appropriate syntax in order to be truth apt? Surely not. First, ellipses might be truth-apt, and yet lack the appropriate syntax.<br><br> Second, sentences governed by force-modifiers might be truth-apt, and yet fail to have the appropriate syntax -- c.f. Lewis (1970). Third -- though this might be construed as a version of the first case -- there are many sentences of the form ADVERB+S which fail the embedding tests, and yet which seem to be truth-apt.<br><br> These are all reasons for disagreeing with the idea that questions about truth-aptness are merely questions about fitness for substitution in the truth-schema. 3 Thesis (4) is special, however. I shall argue that, even if the argument against the package as a whole were to fail, there would nonetheless be good reasons to reject the idea that traditional local realist / anti-realist debates need to be importantly reconceived from the standpoint of one who accepts theses (1)-(3).<br><br> 4 ease of reference threatens an explosion of ontological, ideological and theoretical commitments. If one is concerned to avoid this explosion -- as plausibly one ought to be, and as Wright explicitly is -- then one must make use of strategies which, I claim, are the tools-in- trade of error theorists and/or non-cognitivists. But then it follows that the acceptance of thesis (4) is a mistake -- those techniques which are supposed to be available to rule out commitments to, say, fictional entities (Part 3), can be adapted to make room for error theories and/or non-cognitivisms in domains such as ethics (Part 4).<br><br> So our minimalists should renounce their kind of minimalism -- something in the package of theses must be given up. 1 Much of the vocabulary used to state the theses described above cries out for further explication: e.g., what is truth-aptness? what is a discourse?<br><br> what is a norm of appropriate utterance? what is warranted assertability? what is semantic width?<br><br> and so on. The attempt to provide this clarification often suggests ways in which the theses themselves might be challenged, as the following discussions show: (1) Semantic width : Wright (1992) draws distinctions between different kinds of truth and different kinds of truth predicates. Many commentators -- e.g.<br><br> Williamson (1994) -- have suggested that it would be better to hold that there is just one truth predicate, and one truth, but many different subject matters. In response, Wright (1994:337) suggests an analogy between truth and identity: cThe contention of pluralism about truth is that the instantiation of the truth predicate may be constituted in different ways depending upon the kind of instantiators concerned d. However, as Wright himself concedes, the notion of 8instantiation 9 deployed here requires elucidation; and until such elucidation is supplied, it might be thought that it will remain unclear whether there is now any substantive disagreement between Wright and his critics on this point.<br><br> But perhaps there are other reasons to worry about the thesis of semantic width. Two naturally come to mind. 5 First, if truth may be instantiated in different ways, then presumably reference (to objects and properties) may be so as well.<br><br> But, partly under the influence of Quine, there is considerable resistance to the idea that there are kinds or degrees of existence. Since this will be one of the principle themes of our subsequent discussion, I shall say no more about it here. Second, if one allows that truth can be thicker or thinner, then it seems that one will open the door to the thought that there are several different kinds of falsity which can be paired with truth of a given thickness (and, hence, to the thought that there are several different kinds of negation).<br><br> Suppose we focus on a discourse for which truth is relatively thick. Should the corresponding notion of falsity simply amount to the denial of the presence of thick truth -- or should it amount to assertion of the presence of thick falsity (i.e. should it involve the presupposition that thick truth-aptness is present)?<br><br> Speaking roughly, do we want to allow that there is both internal and external negation, or do we want to insist that there is only one kind of negation? In what follows, I shall try to finesse these difficulties by focussing only on the case of true atomic sentences -- i.e. I shall not give any arguments which turn on the detailed answers which one gives to these questions.<br><br> However, it is worth bearing in mind that the thesis of semantic width may well bring quite a bit of extra ontological and/or ideological and/or theoretical baggage in its train -- and that one might well prefer to leave that baggage behind. (2) Truth : The idea that there is a minimal kind of truth which is to be identified with, or explained by, some kind of psychological or linguistic state, seems to face a severe difficulty in the form of 8blindspot arguments 9. Indeed, even the thought that these psychological or linguistic states provide a necessary condition for truth seems to be demonstrably wrong.<br><br> The general form of argument against the claim that A ¨ -ability is necessary for truth (where ¨ is a psychological or linguistic state verb: believe , assert , utter , conceive ; and A is a 6 modifier: true , correct , warranted , justified ) may be represented as follows: Let [] = Q, but no-one ever A-ly ¨ -s that Q 1. For any R, it is A-ly ¨ -able that R if(f) it is true that R.(Premise for reductio) 2. For any Q, if it is not A-ly ¨ -able that [], then it is not the case that [].(From 1) 3.<br><br> For any Q, it is not A-ly ¨ -able that [].(Premise, supported by sub-argument) 4. (Hence) For any Q, it is not the case that [].(From 2, 3) 5. (Hence) For any Q, if Q, then someone at some time A-ly ¨ -s that Q.(From 4) 6.<br><br> Not: For any Q, if Q, then someone at some time A-ly ¨ -s that Q.(Obvious Premise) 7. (Hence) Not: For any R, it is A-ly ¨ -able that R if(f) it is true that R.(From 1, 5, 6) The sub-argument for Premise 3 is this: In order to A-ly ¨ that [], one must A-ly ¨ that Q. But this contradicts the content of the second conjunct of [], viz.<br><br> that no-one ever A-ly ¨ -s that Q. The conclusion which seems to follow is that truth and A ¨ -ability diverge in extension for any discourse which contains the vocabulary required to state this argument. And this at least casts doubt on -- if it does not outright refute -- minimalist thesis (3).<br><br> Wright objects to this argument: (i) that in no case in which it is plausible to think that A ¨ - ability is necessary for truth does the discourse in question contain the vocabulary needed to state the argument; and (ii) that cwhen we envisage the possibility of different outcomes to the realist / antirealist debates about, for instance, morals, mathematics and comedy, we are concerned with the status of commitments distinctive of those areas of discourse in a sense which -- for moral, mathematical and comic 8P 9 respectively -- someone who asserts 8No-one will ever have a warrant for cP d 9 need not have undertaken a moral, mathematical or comic commitment d. However, it just seems obviously wrong to say that one who says: cNo-one will ever have a warrant for the claim 8Murder is wrong 9 d undertakes no moral commitment. (After all, one who says this must also, on pain of Moorean paradox, refrain from claiming 7 that murder is wrong.) Moreover, it is utterly unclear how discourses are supposed to be individuated, if the result which Wright wants is to be achieved.<br><br> Perhaps, by gerrymandering the notion of discourse, one can maintain that there are discourses for which A ¨ -ability is necessary for truth; but, on any natural conception of discourses (about which more later), it seems most unlikely that there are any discourses of the kind in question. 4 (3) Truth-aptness : Truth-aptness might be thought to be (primarily) a property of any of a number things: sentence-tokens, sentence-types, sentence-tokenings, utterances, statements, propositions -- sets of possible worlds, classes of circumstances of evaluation of sentence- tokenings, abstract structured entities -- thoughts, beliefs, and so on. I shall suppose that truth- aptness -- and, indeed, truth and falsity -- are properties of sentence-tokenings in possible contexts of tokening.<br><br> (Formally, one might represent the bearers of truth-aptness as ordered pairs, the first member of which is a sentence token, and the second member of which is a possible context of tokening.) I assume that the sentence-tokens to be tokened belong to a particular language, i.e. they are not uninterpretted strings of symbols, though it will usually be the case that the interpretation of the token is not independent of the context in which it is tokened. I also assume -- perhaps making the previous assumption redundant -- that a complete specification of a possible context of tokening would require specification of everything relevant to the semantic assessment of any sentence tokened in that context.<br><br> Finally, I shall suppose that the sentence-tokens in questions are in canonical form -- i.e. that their syntactic structure reflects their (deep) logical form. 5 4 Wright has two further, technical objections to the blindspot argument, one concerning the domain of quantification (and facts about indexicality), and one concerning the inference of 5.<br><br> from 4. (which is not intuitionistically acceptable). Since these objections require replies, and since I do not intend to try to provide replies here, I claim merely (i) that the blindspot argument presents a prima facie difficulty for Wright 9s views, and (ii) that his cmost basic concern d about the blindspot argument seems unfounded.<br><br> 5 I take it that this assumption is merely a natural extension of the thought that , e.g., our interest here will be in the semantic reference of singular terms rather than in their speaker reference. A sentence token can be used to (speaker) express all kinds of propositions (and to do all kinds of other things as well); however, corresponding to most sentence-tokenings there will be sentence-tokens whose syntactic form captures the logical form of the (speaker) content of those sentence-tokenings -- and it is these latter sentence-tokens which are the subject of our current interest. 8 Given that truth-aptness is a property of sentence-tokenings in possible contexts of tokening, which such property is it?<br><br> One thought is that it is the property of being either true or false, but not both. There are various well-known reasons why this seems wrong, at least by the lights of the minimalist thesis under consideration: some sentence-tokenings involve ambiguity, some are vague, some are paradoxical, and so on, without violating the two constraints. Consider the empirical paradoxes discussed by Kripke: the sentence-tokenings here have the right syntactic form, and they are subject to considerable discipline; it is merely contingent misfortune which deprives them of truth-value.<br><br> Or consider vagueness: there are various norms which determine when it is, and when it is not, appropriate to make vague assertions -- i.e. there is no question that vague assertions can be suitably disciplined. (Likewise, I say, for ambiguity, though I grant the case is more controversial: there is a norm which enjoins us to avoid ambiguity, but it is defeasible.) 6 A next natural thought is that truth-aptness is the property of being of a kind which is typically or normally either truth or false but not both, e.g.<br><br> when not subject to the problems mentioned in the previous paragraph. But this thought is too vague to be helpful as it stands. And there is really only one way that I can see to develop it, namely, by adverting to the compositional structure of the sentences whose tokens are under consideration.<br><br> Since the idea is pretty obvious, I shall merely outline its development. Moreover, in order to simplify the exposition, I shall merely outline the idea as it would apply if truth-aptness were a property of sentence-types rather than of sentence-tokenings in possible contexts; and I shall also make the simplifying assumption that the logical (grammatical, deep structural) form of the sentence-types under consideration is captured by their surface syntactic form. Suppose that we have some way of identifying predicates and singular terms.<br><br> Some will be 6 Of course, some -- e.g. those who think that vagueness is epistemic, not ontic or semantic -- will dispute the suggestion that some vague assertions lack determinate truth-value. Similarly, some will dispute the suggestion that the empirical paradoxes create truth-value gaps.<br><br> However -- as these example indicate -- there is at least a strong prima facie case to be made for the view that some sentence-tokens which satisfy the minimalist constraints are neither true nor false. 9 structured, some will be primitive, but for our purposes we don 9t need to distinguish. Consider atomic sentences of the form Fa 1 ...a n , where F is a predicate and the a i are singular terms.<br><br> For each predicate F, note that F refers to the property F just in case there is at least one atomic sentence involving F which is either true or false but not both. For each singular term a i , note that a i refers to the object a i just in case there is at least one atomic sentence involving a i which is either true or false but not both. Then -- and here I leave out the details -- note that a sentence is truth-apt just in case it is built up in the right way from predicates all of which refer to properties, singular terms all of which refer to objects, and other bits of linguistic apparatus all of which have appropriate functions for their semantic contents: logical connectives, modal operators, tense operators, and so on.<br><br> 7 (For any of these further bits of linguistic apparatus, note that it has the appropriate function o for its semantic content just in case there is at least one sentence in which it takes widest scope and which is either true or false but not both.) 8 Perhaps this proposal is a little more explicit than it need be. Perhaps all that is needed is the suggestion that we build with components whose use is subject to discipline -- i.e. predicates, singular terms, and operators whose use is governed by norms of appropriate utterance -- and that the end-product should have the right syntactic form.<br><br> However, it seems to me that this 7 It may seem that there is a Quinean option which is overlooked here, viz. to say that a singular term a i refers to the object a i just in case there is at least one atomic sentence involving that term which it true (and [perhaps] that a predicate F refers to the [instantiated?] property F just in case there is at least one atomic sentence involving that predicate which is true); in this case, reference would not be required for truth-aptness, but it would be required for truth. However, as we shall see, this option is not available to our minimalists: for the adoption of this option is just what error theorists and non-cognitivists need, since it makes room for the view that positive nuclear atomic ethical sentences are all false, or all lacking in truth-value (while taking no stand on the question whether more complex expressions involving the use of ethical vocabulary are truth-apt).<br><br> For more about this, see the subsequent discussion of the principle of semantic innocence. 8 Those who think that paradoxical sentences are not truth-apt will insist on a further condition for truth-aptness apart from the compositionality requirement, viz. that there be at least one (possible) context of utterance in which the sentence is either true or false but not both.<br><br> I am inclined to say that paradoxical sentences are truth-apt -- they belong to a kind which is typically either true or false but not both -- though I also recognise the force of the thought that something can only be truth-apt if it is possible for it to be either true or false but not both. Perhaps the best course is to distinguish two senses of truth-aptness. However, for present purposes, I needn 9t worry about this: the paradoxes do not bear directly on the points I wish to make.<br><br> 10 suggestion entails the more detailed account; surely a minimalist of the kind under consideration will agree (i) that any suitably disciplined predicate expresses a property; and (ii) that any suitably disciplined singular term denotes an object -- for then, and only then, can it be platitudinous, as it ought to be, that a suitably disciplined atomic sentence 8Fa 9 is true just in case the object denoted by the suitably disciplined singular term 8a 9 possesses the property expressed by the suitably disciplined predicate 8F 9. 9 This proposal suggests a way out of the difficulties involved in the notion of a discourse: instead of supposing that there is a collection of sentences which are distinctive of a particular kind of discourse, suppose instead that there is primitive vocabulary which belongs to distinctive kinds -- e.g. suppose that there are distinctively ethical predicates, operators, etc.<br><br> Then one can say that a sentence belongs to a given kind of discourse if it contains vocabulary which is of the appropriate kind (thus leaving room for the idea that a single sentence can belong to many different kinds of discourse). 4. Error Theories and Non-Cognitivisms : Given the thesis of variation in semantic width, care must be exercised in the characterisation of positions which are defined in terms of the relevant semantic notions -- e.g.<br><br> cerror theory d and cnon-cognitivism d. Initially, one might have supposed that something like the following characterisations are correct: (1) An error theory about a domain of discourse holds that all positive nuclear atomic sentences which belong to that discourse are false; and (2) non-cognitivism about a domain of discourse holds either (i) that all positive nuclear atomic sentences which belong to the discourse fail to secure truth-aptness; or else (ii) that all positive atomic nuclear sentences which belong to the 9 There are many other kinds of minimalists who accept the thesis about minimal truth- aptness, but who would not accept the suggestions about reference. Quite generally, one might wonder whether the norms of appropriate utterance governing subsentential constituents should be thought of as referential norms, or inferential norms, or substitutional norms, or some package of these, or as something else altogether.<br><br> However, I take it that our minimalists -- i.e. those committed to theses (1)-(4) -- will agree that the norms of appropriate utterance governing subsentential constituents are referential norms, i.e. they will agree to the proposed account of truth-aptness.<br><br> That proponents of other kinds of minimalism -- e.g. that exhibited in Brandom (1984) -- would not thus agree is irrelevant to my current project. 11 discourse do not even aim at truth.<br><br> 10 But brief reflection should prompt one to ask: how thick are the notions of truth and truth-aptness which figure in these characterisations? On the one hand, one might suppose that merely minimal truth and truth-aptness are at issue. So, for example, the claim of the error theorist will be, roughly, that no atomic claims are minimally true (i.e.<br><br> superassertible given the standards of warrant internal to the discourse); and the claim of the non-cognitivist will be, roughly, that no atomic claims are minimally truth-apt (i.e. these claims do not aim for, and/or do not achieve, minimal truth or falsity). Given this understanding, there is some prima facie plausibility to the thought that error theories and non-cognitivisms are untenable in any case in which utterance is suitably disciplined.<br><br> But, of course, it is utterly implausible that historically important error theorists and non-cognitivists took themselves to be committed to the views here considered. 11 On the other hand, one might suppose that some relatively robust truth and truth-aptness -- involving cognitive command, response-independence, wide cosmological role, and perhaps 10 Neither the distinction between positive and negative vocabulary, nor the distinction between nuclear and non-nuclear (or characterising and non-characterising) vocabulary is unproblematic. However, at least over a range of simple cases, it seems that there is considerable convergence in the classifications which theorists are disposed to make.<br><br> (Very roughly: (i) negative vocabulary is vocabulary whose reference is fixed in terms of, or which sustains analytic connections to, expressions which are formulated in terms of negation (and related operators) in canonical theory ; and (ii) non-nuclear vocabulary is vocabulary whose reference is fixed in terms of, or which sustains analytic connections to, semantic vocabulary (truth operators and predicates, existence operators and predicates, modal operators and predicates, etc.).) Little in my argument turns on the use of these distinctions: it would mostly suffice to provide a characterisation by enumeration. 11 Actually, it is not clear that non-cognitivism is untenable even when it is minimal truth that is at issue. The problem is that it is not really clear that, from the minimalist standpoint, non- cognitivism can be adequately characterised in terms of truth.<br><br> On the one hand, it seems wrong to think that the mere presence of truth-value gaps is a sufficient mark -- for the difference between error theories and non-cognitivisms shouldn 9t turn on possibly conventional choices about the treatment of reference failure, etc. But, on the other hand, it seems wrong to think that any area of discourse could have as a characteristic aim the project of aiming for minimal truth -- for minimal truth is all too easy to achieve. Thus, if one shifts to the minimalist framework, it becomes plausible to think that there must be some quite different characterisation of non-cognitivism which is required.<br><br> I shall return to this issue towards the end of the paper. (A similar point applies to error theories as well: given the ease with which minimal truth can be achieved, there must be some other feature in virtue of which error theories should be characterised. I shall not emphasise this point in what follows.) 12 even evidence transcendence -- are at issue.<br><br> In this case, the claim of the error theorist will be, roughly, that no atomic claims are robustly true; and the claim of the non-cognitivist will be, roughly, that no atomic claims are robustly truth-apt (i.e. none aims at, and/or achieves, robust truth). Given this understanding, there is NO prima facie plausibility to the thought that error theories are untenable in any case in which utterances are suitably disciplined; and there is LITTLE plausibility to the thought that non-cognitivisms are untenable in these cases (for why shouldn 9t it be the case that there are discourses which do not even aim at robust truth?).<br><br> Moreover, it is highly plausible that it is precisely these latter kinds of theses to which historically important error theorists and non-cognitivists took themselves to be committed. These considerations suggest that it is quite unclear that these minimalists have even a prima facie entitlement to the view that semantic minimalism forces a re-alignment of philosophical disputes about realism; perhaps all it really does is to require new labels for the same old views. Consider traditional non-cognitivism about ethics.<br><br> The traditional non-cognitivist supposed that ethical utterances do not aim at truth, where the truth in question is supposed to be of a relatively robust kind (involving, say, cognitive command). Under the minimalist regime -- and drawing on the allegedly required distinction between kinds of truth -- there will be the position which holds that ethical sentences do not aim at truth c.c. , where truth c.c.<br><br> is a relatively robust property which requires cognitive command. So the position is the same, and it has the same properties as before -- in what is the advance supposed to consist? Re- arranging the deck-chairs simply doesn 9t alter the substantive debates in any way at all.<br><br> 2 In order to make the case that our minimalist is going to have trouble with reference, I propose to focus on the case of atomic sentence tokens of the form 8Fa 9, under the further pro tem assumption that surface syntactic form reflects logical (semantic, deep structural) form. Moreover, I shall restrict my attention to contexts in which these sentence tokens are superassertible given the standards of warrant which govern their utterance -- i.e. I shall 13 suppose that these sentence tokens comply with relevant standards of discipline and that they meet the conditions for 8unrevisable 9 assertability.<br><br> Finally, I shall suppose that no negative or non-nuclear vocabulary plays any role in fixing the reference of the tokens involved in the sentences -- i.e. singular terms such as 8God 9 are excluded from consideration (under the assumption that the reference of this name is fixed by a description which includes the words 8necessarily existent 9). Examples might include the following: c2 is even d, cSanta Claus has a white beard d, cPaul Keating is Prime Minister d, cHitler was evil d, etc.<br><br> (Positions are much easier to classify for the limited case of 8true 9 atomic sentences which involve no non-nuclear or negative vocabulary -- and this case is all that we need to consider here.) I propose the following scheme of classification of the positions which might be taken: (A) With respect to the singular terms : (i) Meinongianism : Every such singular term token -- or perhaps every such directly referential singular term token -- refers to an object. (a) Noneism : There is no guarantee that a given singular term token will refer to an existing object -- i.e. many such singular term tokens refer to merely non-existing objects; (b) Allism : Every such singular term token refers to an existing object.<br><br> (ii) Someism : It is not the case that every such singular term token -- or perhaps every such directly referential singular term token -- refers to an object; i.e. some (directly referential) singular term tokens of the kind in question fail to refer to objects of any kind. (B) With respect to the predicates : 14 (i) Meinongianism : Every such predicate token expresses an instantiated property.<br><br> (ii) Someism : It is not the case that every such predicate token expresses an instantiated property; i.e. some such predicates fail to express instantiated properties. The question to be asked is: which of these positions are the ones which are (or should) be occupied by our minimalists?<br><br> Consider the case of singular terms. Prima facie , it seems very plausible to suppose that Meingongianism is the natural position for someone committed to theses (1)-(4). After all, it seems more or less platitudinous to say that, if a sentence 8Fa 9 of the kind in question is true, then the object referred to or denoted by the singular term 8a 9 has the property expressed by the predicate 8F 9.<br><br> (Remember that I have insisted that logical form, etc. is reflected in surface form.) And it also seems that our minimalist does suppose that all of the sentences in question are true, since -- ex hypothesi -- she supposes that they are superassertible given the standards of the discourse to which they belong. Of course, since our minimalists distinguish between kinds of truth -- or, at least, they suppose that truth varies in width -- they may very well distinguish between kinds of being as well.<br><br> Moreover, Wright himself seems to have good reason to embrace this consequence of his views, since parallel to his various 8cruces 9 for truth in local realism / anti-realism debates -- cognitive command, response independence, width of cosmological role, evidence transcendence -- there are corresponding 8cruces 9 for existence of objects in local realism / anti-realism debates -- formal requirements of consistency and completeness, response independence, width of cosmological role and evidence transcendence. One of the loci of debate about the existence of objects concerns their 8mind-independence 9; some of the 8cruces 9 in question can be construed as ways of spelling out this idea. (Consider, e.g., Brentano 9s 8intentional objects 9: there is some plausibility to the thought that these fail to have wide cosmological role.) 12 12 The discussion in this paragraph is not meant to suggest that (the formulation of) any of 15 However, this feature of Wright 9s view -- which is also characteristic of typical Meingongian views -- is irrelevant to the classification which I have given.<br><br> Noneists and allists differ on the question whether existence requires a relatively robust degree of being, not on the question whether there are degrees of being. While one might think that an investment in degrees of being is obnoxious, to pursue that line of thought would be to return to the kinds of objections to the thesis of semantic width mentioned in the previous part of this paper. Should our minimalists be noneists?<br><br> Perhaps. However, we know that Wright is not a noneist, for he thinks that his minimalism alone commits him to arithmetical Platonism, i.e. to the existence of numbers.<br><br> 13 Moreover, if our minimalists are noneists, then we are clearly not being offered anything new: the position has already been developed and defended at great length by Sylvan (1980) and others. Of course, there may be nothing wrong with old wine in new bottles; however, I shall later suggest some other reasons for being dissatisfied with this option. And I note that one should not make the mistake of supposing that the new bottles improve the wine.<br><br> Wright 9s 8cruces 9 is unproblematic. For instance, given Wright 9s account of cognitive command -- roughly, that a discourse exerts cognitive command iff it is a priori that difference of opinion formulated within the discourse, unless excusable as a result of vagueness in a disputed statement, or in the standards of acceptability, or variation in personal evidence threshholds, will involve something which may properly be regarded as cognitive shortcoming 3 it is unclear to me (i) whether any discourse could fail to conform to it; and (ii) whether any discourse could be required to conform to it. The problem is what is to count as 8cognitive shortcoming 9.<br><br> If it has to do with reasonable procedures of belief revision, then it seems to me that the role of chains of testimony over long periods of time could bring it about for almost any discourse that there is divergence in opinion even though there is no- one in the chains who has violated any of the canons of belief revision. On the other hand, if it has to do with the refusal to tolerate contradictions, then -- following Williamson (1984) -- I don 9t see how anything which deserves to be called a discourse could fail to observe it. Since nothing in the sequel will turn on this, I shall not pursue these worries -- or worries raised by the other 8cruces 9 -- here.<br><br> 13 Divers and Miller (forthcoming) suggest that this point shows that Wright will be hard pressed to avoid allism. Perhaps so. But Wright could hold on to his arithmetical Platonism if he found some other reason for espousing it; or he could decide to give up on his arithmetical Platonism; or he could retreat to a noneist position (i.e.<br><br> he could accept that numbers are objects while denying that they exist). Thus, this consideration alone does not suffice to undermine the kind of minimalism under examination. 16 Should our minimalists be allists?<br><br> Surely not; allism is ontologically (existentially) profligate to an extraordinarily objectionable degree. Moreover, we know that Wright is not an allist, for he thinks that certain ontological questions -- about the existence of referents for singular terms -- are not trivial in the way that they must be for allists. Should our minimalists be someists (as it seems they ought to be, and as Wright wishes to be)?<br><br> Well, if so, then there are some hard cases for them to confront. Can they give a plausible account of fictions, radically false beliefs and theories, assertions proper to conflicting religions, assertions proper to typical debates about sporting teams, and so on, while at the same time continuing to maintain that minimalism provides principled objections to non- cogntivisms and error theories in, say, ethics? Can they give a principled account of sentences like these -- Santa Claus has a white beard ; The air in the bottle is dephlogisticated ; God sent his only son , Richmond is the best team in the league -- which does not use machinery which could be adapted to give an error-theoretic or non-cogntivist account of ethical sentences?<br><br> I doubt it. In order to explain why, I shall consider the case of utterances about fictional characters. (If you think that utterances about fictional characters should not be given special treatment, then choose some other case which you think does need such treatment.<br><br> Unless you are a Meingongian, you will think that there is some such case.) I shall suggest that, in order to deny that a fictional name 8a 9 refers -- in the circumstances envisaged earlier -- one will need to adopt an error-theoretic or non-cognitivist strategy. But, if this is right, then it seems to follow -- and I will argue that it does follow -- that our minimalists have no principled objection to error theories and/or non-cogntivisms in ethics. 3 Consider, then, the case of utterances about fictional characters -- e.g.<br><br> cWatson lives at 221B Baker St. d, cWatson is the most interesting character in the Conan Doyle novels d, etc. Two 17 broad kinds of strategies plausibly suggest themselves for those who wish to avoid commitment to existing fictional characters, viz: (i) denial that there is genuine, full-blooded assertion and/or denial that there really are norms of appropriate utterance; and (ii) denial that the genuine logical form of the sentences uttered corresponds to their surface form. I shall consider these strategies in turn.<br><br> (Of course, one might deploy both of these strategies together; but that fact will offer no comfort to our minimalists.) 1. No Genuine Assertion : One standard non-cognitivist ploy, when confronted with a target discourse for which non-cognitivism is deemed appropriate, is to deny that there is any genuine assertion of the sentences in the discourse. This denial can take one of two forms.<br><br> First , one can maintain that there is only quasi-assertion -- e.g., in the case of fiction, that there is only the pretence of assertion -- and that there are only quasi-norms of appropriate utterance -- e.g. in the case of fiction, that there is only the pretence that there are norms of appropriate utterance. Second , one can draw a distinction between loose and strict speech, and then insist that there is only loose assertion (conformity with loose norms of utterance) but no strict assertion (no conformity with strict norms of utterance).<br><br> In the case of fiction one will hold that, speaking loosely, Santa Claus does have a white beard (for, after all, that is what the relevant tradition maintains); but, speaking strictly, it isn 9t the case that Santa Claus has a white beard (since, of course, there is no Santa Claus). Of course, the adoption of either one of these strategies seems to fit uncomfortably with the minimalist theses: for, if thesis (3) is meant to allow an apparently non-minimalist distinction between genuine and non-genuine assertion -- or between loose assertions and strict assertions -- then it is clear that thesis (4) is completely lacking in justification: it is characteristic of error theorists and non-cognitivists in ethics to hold that (some) ethical claims are not genuinely and strictly asserted. 2.<br><br> Hidden Logical Form : Another standard non-cognitivist ploy, when confronted with a target discourse for which non-cognitivism is deemed appropriate, is to deny that genuine 18 logical form corresponds to surface syntactic form. Again, this strategy can take one of two forms. First , one might insist that there are implicit prophylactic operators which provide protection against unwanted commitments: e.g.<br><br> caccording to the fiction d, caccording to the theory d, etc. (According to the Santa Claus fiction, Santa Claus does have a white beard; and, according to the moral fiction, murder is wrong.) Second , one might insist that the problematic vocabulary disappears under paraphrase -- i.e. that apparently ontologically or ideologically or theoretically committing vocabulary disappears under paraphrase or translation into deep structure, etc.<br><br> (Appropriate examples for the case of fiction aren 9t immediately forthcoming: after all, Russellian and Quinean treatments of proper names seem to give paraphrases of cSanta Claus has a white beard d which do not eliminate the problematic commitments. However, metalinguistic analyses would be a case in point.) Once again, the adoption of these strategies doesn 9t seem to fit well with the minimalist theses: for, if the csyntactic tests d adverted to in (3) refer to cdeep syntax d -- as opposed to the surface syntactic forms which one would naturally have supposed to be at issue -- then there is plenty of room for error theorists and non-cognitivists to manouevre in the case of ethics (c.f. the concluding section of this paper).<br><br> Moreover, it isn 9t clear that these strategies do all that the minimalist requires them to do. It is true that the implicit prophylactic operator approach provides true sentences in one-one correspondence with the allegedly false sentences of the fictions. But it doesn 9t seem at all plausible to suppose that the utterance of these true sentences is always the only (or even the most important) aim of one who produces the original sentence tokens.<br><br> When I am recounting the Santa Claus story to my children, it seems clear that there is no sense in which I mean to assert the true companion sentences. Indeed -- and this is what motivates a non-cognitivist or error-theoretical account of the fictional sentences in question -- it seems that there is no truth which my utterances are designed to communicate (though, of course, I do try to ensure that the sentences which I utter conform to the Santa Claus story). 14 14 The considerations raised here point to a consideration which is often undeservedly ignored, viz.<br><br> that one might want to say different things about different uses of tokens of the same type. Consider 8Santa Claus has a white beard 9. Perhaps it is right to think that the use 19 3.<br><br> Wright On Fiction : Wright himself espouses an error-theoretic treatment of fiction: c[W]hat is distinctive of fiction [is that] to have warrant [for the assertion of a fictional statement] is to have no reason whatever to regard such a statement as true d (p.328). But one should ask: What reason do we have for making fictional assertions in those typical circumstances in which we know that they are false? Isn 9t there supposed to be an internal connection between the aim of assertion and truth?<br><br> Why isn 9t it correct to interrupt someone who starts out cOnce upon a time ... d with cFALSE! d? (After all, other things being equal, we don 9t knowingly allow falsehoods to pass unchallenged in other domains of discourse.) Isn 9t there great pressure here to concede that one isn 9t REALLY making assertions when, e.g., one tells a story? Moreover, even if one waives the previous difficulty, what is the reason for thinking that the behaviour in question is characteristic of fiction?<br><br> Why can 9t an error-theorist about ethics claim exactly the same status for ethical discourse. (We don 9t usually interrupt ethical statements with cFALSE! d But this could be for reasons similar to those in the case of fiction.) Perhaps -- as Tim Williamson suggested to me -- the idea is that our use of terms like creally d shows that the error-theoretical perspective is internal to fictional discourse in a way that it is not internal to ethical discourse. But consider the following pair of dialogues: (A) cSanta Claus has a black beard! d cReally? d cNo, not really; he has a white beard. d cSo Santa Claus REALLY has a white beard? d cNo, not really.<br><br> There is no Santa Claus. All that 9s really true is that, according to the Santa Claus fiction, Santa Claus has a white beard. d of this token in telling a story is not truth-apt -- i.e. that a non-cognitivist analysis is appropriate -- even though other uses of it would simply be false.<br><br> Compare with 8Murder is wrong 9. Perhaps there are uses of it which should be given a non-cognitivist analysis (their primary function IS simply to express a certain kind of disapproval), even though there are other uses of the token which are simply false. (Compare with tokens of the sentence: cYou will shut the door. d In context, this can be a command, a question, a prediction, and so on -- depending upon factors like stress, intonation, etc.) In keeping with usual practice, I shall not worry about these considerations in the remainder of this paper.<br><br> 20 (B) cMurder is good! d cReally? d cNo, not really; murder is wrong. d cSo murder is REALLY wrong. d cNo, not really. There are no ethical truths. All that 9s really true is that, according to the moral fiction, murder is wrong -- oh, and that I strongly disapprove of murder. d I don 9t see that there is any internal difference between the use of creally d in these dialogues; and I don 9t see how there can be any non-question-begging argument from these kinds of considerations for the view that error-theoretical perspectives on ethical discourse are mistaken.<br><br> 15 (Of course, many of us are NOT error theorists about ethics; and most of us are NOT realists about Santa Claus. But these statistical considerations are unimportant; what matters are the reasons available for adopting these positions. An error theory about 8God 9 does not become untenable just because everyone else believes.) 16 The immediate conclusion which seems to follow from these considerations is that Wright 9s favoured treatment of fiction does not fit with the central package of minimalist theses.<br><br> However, it also seems to follow that the only treatment of fiction which is compatible with that package of theses is Meinongian -- and that seems to be bad news for our minimalists. Two objections may suggest themselves at this point. First -- a point which some readers will be dying to make -- there are semantic theories which will deny that minimalists must be Meinongians, on the grounds that minimalists can avail themselves of substitutional quantification, free logic, and the like.<br><br> Second , it may appear that it is a point in favour of the 15 Part of the point being made here is that one function which creally d can have is as a pointer to formulations in canonical notation, i.e. to notation which is not systematically misleading, and in which surface syntactic form corresponds to deep logical form. ( cDo you REALLY claim that murder is wrong? d cWell, sort of.<br><br> I do strongly disapprove of murder; but, of course, I don 9t think that there is an objective fact of the matter. d) 16 Just for the record, I think that it is worth remarking that it may well be the case that some philosophers are greatly disposed to underestimate how many people are either (i) error theorists and/or non-cognitivists about ethics (nihilists and expressivists); or else (ii) theists who accept the conditional claim that if there were no God, then there would be no morality (i.e. people who are not error theorists and/or non-cognitivists about ethics, but only in virtue of their acceptance of other claims about which one might well wish to be an error theorist or non--cognitivist). Perhaps this reflects on the company I keep, but most non-philosophers I meet are expressivists -- and they are disposed to say just the kinds of things that the second participants in my little dialogues say.<br><br> Their use of 8REALLY 9 simply doesn 9t bear out the claim that fictional discourse is relevantly different from ethical discourse. 21 minimalist view -- and against the someist alternatives -- that it involves no denial of what we might call cthe principle of semantic innocence d. I shall consider these objections in turn.<br><br> 4. Substitutional Quantification, Etc : I have characterised Meinongianism about singular term tokens as the view that all nuclear singular term tokens which appear in superassertible atomic sentences refer to objects. However, it might be thought that this characterisation wrongly suggests that substitutional quantification involves commitment to Meinongianism -- i.e.<br><br> it might be thought that recourse to some kind of substitutional account of the quantifiers will be compatible with our minimalist theses. In particular, one might suppose that the earlier account -- developed in the discussion of truth-aptness -- actually provides an analysis or definition of reference (to objects and properties) in substitutionalist terms. I hope that it is clear that this objection is misconceived.<br><br> If a friend of substitutional quantification does suppose that my account provides an analysis or definition of reference (to objects and properties) in substitutionalist terms, then clearly that person is a Meinongian (by my characterisation). If, on the other hand, the friend of substitutional quantification is not Meinongian -- i.e. if she supposes that some nuclear singular term tokens which appear in superassertible atomic sentences do not refer to objects and/or that some nuclear predicate tokens which appear in superassertible atomic sentences do not refer to instantiated proprties - - then it seems clear that the argument deployed against Wright will go through.<br><br> After all, in such a case, there will be a restricted quantifier which takes as values only the referring terms and instantiated predicates -- and the debates about fiction and ethics can be carried out with respect to it. (Perhaps it is worth reminding the reader that, on my characterisation, one can be a Meinongian even if one does not think that there are degrees of being, but that the minimalists under discussion independently acccept that there are degrees of being.) 5. Semantic Innocence : Suppose one opts for an error-theoretic or non-cognitivist treatment of fiction.<br><br> Then -- by the arguments already given -- it seems that one will be committed either to the view that 8Santa Claus has a white beard 9 has no truth-value, or else to the view 22 that 8Santa Claus has a white beard 9 is false, because the term 8Santa Claus 9 fails to refer (perhaps because it is not subject to norms of appropriate strict utterance). But how then are we to understand the claim that the sentence 8According to the fiction, Santa Claus has a white beard 9 is true? If we think that the truth-value of the whole sentence is recursively determined by the semantic values of its component vocabulary, then we are committed to the idea that there is a semantic gap associated with the name 8Santa Claus 9.<br><br> But won 9t this semantic gap infect all sentences in which the name 8Santa Claus 9 is used? One hopes not. Consider the sentence 8He said that the very sentence he was then uttering was false 9.<br><br> It is very plausible to think that this sentence would be true of a man who had just uttered a liar sentence, even though it contains a paradoxical component which seems to engender a truth-value gap for that component. Or consider any examples which might be offered by defenders of 8implicit operator 9 accounts of fiction, etc. Unless one accepts some kind of Meinongian theory, one will want to say that claims about what happens according to a fiction, or what is true according to a theory which contains non-referring terms, can be true, even though there is a gap in the semantic content of the sentences used to express those claims.<br><br> One option is to go metalinguistic, i.e. to claim that operators which can yield truth-valued output despite receiving input which contains semantic gaps actually treat the input which they receive as if it were encased in quotation marks. So, for example, 8He said that the very sentence he was then uttering was false 9 would be true only because of the truth of the sentence 8He said: cThis very sentence is false d. 9 Similarly, 8According to the fiction, Santa Claus has a white beard 9 would be true only because of the truth of the sentence 8According to the fiction: cSanta Claus has a white beard d. 9 However, there are well-known problems with this kind of analysis: e.g.<br><br> questions about translation, quantifying-in, and so on. The sentence 8There is somewhere such that, according to the fiction, Santa Claus lives there 9 is surely true; but the simple quotational analysis provides gibberish rather than a corresponding 23 truth. 17 While there is reason to reject the metalinguistic analysis, it seems clear that it embodies the correct strategy, viz.<br><br> to look for some other semantic value for the embedding operators to operate upon. Given that 8Santa Claus 9 is to make some contribution to the content of the sentence 8According to the fiction, Santa Claus has a white beard 9, it must be the case that 8Santa Claus 9 has some semantic value other than a referent (for, of course, given that were are not Meinongian, we do not suppose that it has one of those). Following Frege, we could call this semantic value a 8sense 9 -- though we would surely not want to adopt Frege 9s account of what senses are.<br><br> Given the way this strategy has been developed, it should be clear that it involves denial of what might be called 8the principle of semantic innocence 9, namely, the idea that the contribution which vocabulary makes to semantic content does not vary as the surrounding vocabulary in the sentence varies. The idea is not to give 8Santa Claus 9 some further semantic value which might make the sentence 8Santa Claus has a white beard 9 come out true; and yet we do want to give 8Santa Claus 9 some further semantic value which will make the sentence 8According to the fiction, Santa Claus has a white beard 9 come out true. I don 9t propose to pursue the details of this story here, and nor do I propose to sing its virtues.<br><br> 18 However, I do want to point out that cases such as that of fiction do lend considerable support to the thought that one ought to reject the principle of semantic innocence. Indeed, it seems that it is only Meinongians who ought to be enamoured of the principle: anyone who accepts that there can be genuinely empty names, say, will face the problem which the denial of semantic innocence is naturally fitted to solve. (This last remark 17 Of course, there are more sophisticated metalinguistic analyses which can handle this example.<br><br> However, I suspect that one can always find tricky counter-examples to defeat metalinguistic analyses -- and I shall proceed under the assumption that this suspicion is correct. 18 See Oppy (1992) for some independent arguments in favour of rejection of semantic innocence. 24 should help to explain my earlier claim that our minimalists cannot adopt the Quinean suggestion that positive nuclear atomic sentences which involve non-denoting singular terms and predicates are all false or lacking in truth-value.<br><br> For, of course, there is a parallel strategy open to the ethical error theorist which will allow her to hold that positive nuclear atomic ethical sentences are all false (because the predicates involved do not pick out instantiated properties); and there is a similar strategy available to the ethical non-cognitivist which will allow her to hold that positive nuclear atomic ethical sentences are neither true nor false (because the ethical predicates do not express properties) -- and yet denial of the principle of semantic innocence will also allow these error theorists and non-cognitivists to maintain that more complex expressions involving ethical terms are truth-apt. Of course, it is not standard for non-cognitivists to hold that people can have ethical beliefs; however, the Fregean strategy offers a way to make sense of the idea that one can believe that murder is wrong even though there is no property of wrongness. 19 And if one wants a more orthodox non-cognitivism, then one can hold that belief-sentences fail to be truth-apt if they embed otherwise naked predicates which fail to express properties, while also holding that there are other operators which are not thus constrained.<br><br> 20 ) 4 The obvious conclusion to draw from the discussion of the preceding section is that our minimalists -- i.e. those who wish to espouse all four of the theses used to characterise their position -- had better be Meinongians: i.e. they should suppose that every singular term whose use is subject to appropriate norms of utterance, and which is used in at least one superassertible atomic sentence, refers to an object; and that every predicate whose use is 19 If these claims seem to you to play too fast and loose with the notion of 8non-cognitivism 9, then please just pass them by.<br><br> I have already noted that there is a large grey area about which there might be thought to be merely a terminological choice involved in the decision to apply either of the labels 8error theory 9 or 8non-cognitivism 9. 20 Another alternative is to distinguish two kinds of belief constructions, or two interpretations of belief-sentences, say 8de dicto 9 and 8de re 9. Read 8de dicto 9, belief-ascriptions embedding ethical claims are truth-apt; read 8de re 9 they are not.<br><br> Something like this could be construed as a variety of non-cognitivism (non-factualism). 25 subject to appropriate norms of utterance, and which is used in at least one superassertible atomic sentence, expresses an instantiated property. Two obvious questions which now arise are these: Should one be a Meinongian?<br><br> and Can Meinongians be non-cognitivists and/or error-theorists about given areas of discourse? I shall take these questions in turn. As I see it, the most important issue which divides Meinongians from others concerns the question whether one thinks that every disciplined discourse needs to be 8taken at face value 9 (which is the Meinongian view), or whether one thinks that there are privileged discourses which can be used to replace, or reduce, or somehow explain away other discourses (or at least explain away the apparent ontological and/or ideological and/or theoretical commitments of those discourses).<br><br> In my view, the Meinongian position is massively implausible. First, it fails to take account of widespread, though admittedly controversial, intuitions about the primacy of certain domains of discourse, e.g. physical science.<br><br> And, second, it fails to mesh with the intuitive appeal of the commitment-avoiding strategies which were discussed in connection with the case of fiction, in the sense that it imputes redundant ontological and/or ideological and/or theoretical commitments wherever one is disposed to think that these strategies are appropriate. Consider the sentences 8Santa Claus has a white beard 9 and 8According to the Santa Claus fiction, Santa Claus has a white beard 9. A Meinongian thinks that we are required to acccept both sentences, strictly understood, whereas a non-Meinongian may think that only acceptance of the second is required.<br><br> Of course, our Meinongian may feel that, given acceptance of one of these sentences, acceptance of the other sentence is cheap -- but, even so, it is clear that, from the standpoint of acceptance of sentences, the Meingongian package costs more. (There is some ontological and/or ideological and/or theoretical cost involved.) So where does the Meinongian think that there is saving to be made? Only in the (allegedly) simpler semantic principles and simpler semantic theory which the Meingongian accepts.<br><br> The Meinongian purports to have a simpler account of semantic values -- no senses -- and a simpler account of compositionality -- semantic innocence; but that is all he has to set against the cost of a massive inflation of acceptable basic sentences (hence a massive inflation in ontological and/or ideological and/or theoretical commitments). Consequently, it seems 26 clear to me that the non-Meinongian strategies will often be worth pursuing. 21 Suppose, then, that one opts for the Meinongian option.<br><br> Can one be a non-cognitivist or an error-theorist about a target discourse? Well, initially, it may seem clear that the answer is cNo! d After all, consider any nuclear predicate 8F 9 whose use is subject to appropriate discipline. Then, by Meinongian lights, the atomic sentence 8The F is F 9 is true.<br><br> So it isn 9t true that any atomic sentence involving 8F 9 is false -- and so our Meinongian cannot give an error theory, let alone a non-cognitivist account, for 8F 9. (This argument may seem a bit quick: who says that the atomic sentence 8The F is F 9 will be suitably disciplined, let alone superassertible? Meinong himself.<br><br> After all, if one is going to allow that 8Santa Claus has a white beard 9 is true (because superassertible according to the Santa Claus story), then presumably 8The F is F 9 must also be allowed to be true (because superassertible according to the theory which Meinong develops). Certainly, it IS true that Meinong believed that, for any nuclear predicate 8F 9, the sentence 8The F is F 9 is true. And there are now plenty of followers of Meinong around who also accept Mally 9s principle, and whose speech is clearly subject to all kinds of discipline.<br><br> It would be very odd to suppose that claims about impossibilia in works of fiction are true even though claims about impossibilia in Meinong 9s work are not.) 22 The argument just given applies only to nuclear vocabulary -- and with good reason. As Meinong himself well knew, there are many non-nuclear predicates and singular terms for which one will wish to make use of the strategies which I have claimed are the tools in trade of error-theorists and non-cognitivists. Consider, for example, a singular term whose reference is fixed by a non-nuclear description -- e.g.<br><br> 8God\x

less