Meaning and Deeds:  Resurrecting Ascriptivism
Merrill Ring
In the 1960's Peter Geach and John Searle produced related critiques of what Geach called "a pattern of philosophizing" and Searle "a pattern of analysis" a pattern paradigmatically exemplified in Richard Hare's claims about 'good' and Peter Strawson's account of 'true'.[1]  Those critiques, held by both Geach and Searle to "refute" the pattern ('Ascriptivism', p 250; Speech Acts , p. 139) and taken by subsequent philosophers to be completely successful in that, have become a powerful weapon in contemporary philosophy.  They are used in the examination of particular concepts: I will often refer to Alexander Miller's Introduction to Contemporary Metaethics where Geach's objections are used repeatedly.[2]  Moreover, the Geach-Searle objections to that pattern are the foundation for the widely accepted exclusion of pragmatics from semantics.[3]
That pattern of analysis or of philosophizing was not given a name either then or now.  Geach refers to "theories of non-descriptive performance" ('Ascriptivism', p. 252) - that has some plausibility as a name for the pattern but it is too vague.  Searle called the pattern "the speech act fallacy". However,that is not suitable on two grounds:  it fails to distinguish speech act analyses that Searle would approve of ("this objection does not hold agains the speech act analysis of the performative verbs", Speech Acts,  p. 138) from those he thinks are fallacious and, secondly, clearly it is not a name that the defenders of such a pattern would accept for their own views.  The proponents of the two domain semantics-pragmatics distinction might call such a pattern 'a confusion between semantics and pragmatics' but, again defenders of the view would not accept that designation of what they are up to.
Geach was particularly interested in the claims made by H.L.A. Hart in his 'The Ascription of Responsibility and Rights'[4], claims about such notions as being voluntary or done on purpose.  Geach labeled the particular exemplification of the pattern found in Hart "ascriptivism".  That label has become the standard way of referring to the pattern's application to such terms as 'done on purpose'.  In the absence of a generally accepted name, I shall extend Geach's name 'ascriptivism' to the entire pattern.  Others have already done so.[5]  It must be remembered that the name 'ascriptivism' is being re-oriented here. 
That pattern of philosophizing, which was not fully set out by Geach or Searle, or, so far as I know, by anyone, has the following paradigmatic shape.[6]  (1) Some "philosophically exciting term T ('Assertion', p. 266) is denied, contrary to the philosophical tradtion, to be the name of a property, feature, attribute; (2) hence, calling something T is not a report that the thing has the property T, is not a description of the thing; (3) calling something T, saying 'That is T', must therefore be doing something other than reporting or describing; (4) whatever is thereby done is an important part of the meaning of T. 
The critics, Geach, Searle and their heirs, direct their specific objections to item (4), the idea that whatever is done in saying 'That is T' must be involved in the meaning of T.  The earlier steps in the pattern, especially the denial that the term in question is the name of a property, are typically ignored by critics of ascriptivism.
My aim here is first to argue that the objections found Geach and Searle do not show that pattern of analysis, ascriptivism, to be mistaken, do not refute it.  Secondly, and in light of the rejection of the Geach-Searle position. I shall defend ascriptivism, although in a version both purged and expanded.  The success of that project means that keeping semantics free of pragmatics is not possible.
Before getting down to work, there are a few necessary program notes.  Searle speaks of the "classical speech act analysts".  Since the 1950's are only jarringly thought of as a classical period, I shall speak of the mid-(20th)-century analysts.  In doing so, I shall be retaining the idea of analysis although what was being done mid-century does not fit what G.E. Moore or Russell, the founders of analytic philosophy, would have thought of as analysis.  Analysis, an account of meaning, as understood both by Searle and here, will have more in common with Ryle's notion of logical geography and Strawson's connective analysis.[7]
It will not be part of my project to explain and defend by name and citation those practitioners of mid-century analysis who were engaged in the ascriptivist pattern of thinking. I shall be referring only to the pattern without detailed concern for particular figures in its development and use.
Strictly the argument here will focus on the analysis of 'good'. (In doing that I am not even committed to holding that all the other mid-century cases are in fact correct.) Occasionally I shall make some reference to 'true' but only as a reminder that the argument is intended to have wider application than to a single concept. The reasons for selecting 'good' as the test case is that it is central to the criticism in both Geach (though Geach prefers to talk of 'bad') and Searle; it is also the notion in contemporary thought that is most often and most vigorously the target of critics of ascriptivism.
About 'good' the mid-century analysis (derived chiefly from Richard Hare) held that the claim 'That is good' is a commendation of the item referred to. As a result, Geach and Searle treat commending as the act performed when one calls something good. In light of the acceptance of that terminology by supporters and critics alike, I shall continue to speak of commendation. But it should be pointed out that the background idea is that 'good' is an evaluative word (notice the talk of 'evaluative meaning') and that it is used to make a positive evaluation (or assessment.) Thus it is not necessary speak of commending as the deed performed when goodness is ascribed to a thing, but it makes philosophical life easier to do so.
The ascriptivist pattern as I set it out earlier can now be exhibited in terms of 'good'. (1) 'Good' is, contrary to the traditional view, not the name of, does not refer to, a property, attribute, character, feature, of a thing; (2) hence to call something good is not to report that it has the property of goodness, is not to describe the thing; (3) what we are doing when we call something good is commending it; (4) the notion of commendation is central to the meaning of 'good'.
It is important to notice that no party to the dispute denies that to say 'That is good' is to commend the item. Geach holds that such an act is of no concern to philosophers. "Of course an asserted proposition in which 'bad' is predicated may be called an act of condemnation. But that is of no philosophical interest." ['Assertion', p. 269] Searle, on the other hand, thinks it not only true but of great philosophical interest that calling something good is commending it – the mistake as he sees it lies in the conclusion drawn from noticing the commendation. "Calling something good is characteristically praising, commending or recommending it. But it is a fallacy to infer from this that the meaning of 'good' is explained by saying that it is used to perform the act of commendation." (Speech Acts, p. 139) Those who defend the strong semantics-pragmatics distinction also accept that calling something good is commending it – they, developing the Geach-Searle view, locate the source of that fact in some other feature of language use than the semantics of 'good'.
The argument in both Geach and in Searle proceeds by noting a distinction between calling something good, i.e. saying 'That is good' (treated here as the paradigmatic form of calling something good: 'It is a good X' is assumed to be covered by the discussion), and other grammatical constructions (other linguistic activities) in which the term 'good' can occur. Those other occurrences include 'That is good' in either the antecedent or consequent of a conditional statement, in the past tense ('That was good'), in a disjunction ('Either that is good or I'm a monkey's uncle'), in a supposition, a question, an expression of curiosity ('I wonder whether that is good') and others. It would be too tedious to try to make reference to all of those other constructions as we move through the discussion. So I shall focus (almost) solely upon the construction that is most relied upon in these discussions: the occurrence of 'good' in the antecedent of a hypothetical. So far as I can tell whatever can be said about 'good' in 'If that is good, then...' applies to all the other various constructions.
There are several matters, some major, about which I shall be departing significantly from the mid-century versions of the pattern. Purging ascriptivism of the first of these might appear shocking as it is often taken to be a defining feature of the pattern. I shall be rejecting the idea that claiming something to be good, saying 'That is good', does not have a truth-value, that ascriptivism is a non-cognitivist position. Deeply influenced by Austin's realization that to say 'I promise' is to promise and by the recognition that, despite its grammatical form, 'I promise...' has no truth-value but rather is a doing of something, an act, the original ascriptivists assumed that if saying 'That is good' is commending something, it cannot have a truth-value. But as Austin himself very soon came to realize doings, illocutionary acts, are all over the place - in saying, e.g., 'That is yellow', we are describing an object, we are engaged in the act of describing. However, it does not follow that 'That is yellow' is neither true nor false. Hence only the narrower class of speech-acts called 'performatives', of which 'I promise' is the model, are those without a truth-value. Both (for example) 'That is yellow' and 'That is good', no matter how different in logical character otherwise, cannot be excluded on speech act grounds from having a truth-value. So someone who says 'That is good' will be right or wrong, will either hit or miss the truth, no matter how different from 'That is yellow' the means of establishing its truth (or falsity) are. Consequently, the ascriptivist pattern is not a non-cognitivist position. There have been many powerful arguments directed against this feature of mid-century ascriptivism, reminding us that we do treat claims to the goodness of things as having truth-values – I do not cite them here, preferring to point out that its non-cognitivism was a mistake derived from assimilation to performatives.[8] 
A second major item to be purged from the earlier version of ascriptivism is the sharp distinction made then between evaluative and descriptive meaning and the associated claim that the descriptive meaning is dependent upon the evaluative. While there is a truth distorted in that bifurcation of two kinds of meaning, both the bifurcation and the dependence thesis are not part of a re-modeled ascriptivism. Once again, a good deal of powerful criticism has been directed at this set of ideas.
The rules of the game having been set out, it is time for the substance. The core of the critics' position is a problem and a challenge. The problem: the ascriptivist pattern of analysis holds that 'That is good' is, by meaning, a commendation; but obviously when that same phrase is located in 'If' clause there is no commendation; yet 'good' means the same in both constructions. The challenge: How is that possible?
Geach and Searle and those who make the consequent semantics-pragmatics distinction agree that it is not possible – and hence hold that the commending cannot be involved in the meaning of 'good' and must be allocated to some other feature of language than its semantics. On the other hand, whoever defends the pattern must show how it is possible that the word means the same in the different constructions and yet in only one of them does commendation occur. My complaint against the mid-century analysts is that they did not so much as attempt to show how that is possible. I shall be required to rectify that shortcoming in reviving the position.
Searle introduces a test for the success of any analysis, for any account of the meaning of a term. "Now, there is a condition of adequacy which any analysis of the meaning of a word must meet.... Any analysis of the meaning of a word (or morpheme) must be consistent with the fact that the same word (or morpheme) can mean the same thing in all the grammatically different kinds of sentences in which it can occur. Syntactic transformations of sentences do not necessarily enforce changes of meaning on the component words or morphemes of those sentences." (Speech Acts, p. 137)
Setting aside the "necessarily" (that shifts the meaning of what he intends), Searle's test is 'A word is not changed in meaning by being located in different syntactical settings'. Searle's use of that test presumes that if ascriptivism were correct, 'good' would change meaning in the syntactical change from 'That is good' to 'If that is good'. That is the reverse of the problem as I set it out above: there the assumption is that the word does mean the same and a problem thereby ensues. While there are serious troubles with Searle's test – are we more certain that the test is correct than what we say about meaning in particular cases? – let it stand since Searle does make use of it.
Given the problem situation, (1) what are the considerations offered by Geach and Searle to show that the ascriptivist pattern fails and (2) what efforts do they make to show that some other analysis, some other account of meaning, can succeed where the pattern fails? Both questions raise matters that are involved in updating the pattern. For reasons having to do with the aims of this paper, let me begin with the second question, an issue that may not seem the best place to start but pursuit of which reveals important matters.
Geach tries to show that even if ascriptivism's account of meaning is rejected, it is possible to provide an account that does not run afoul of the problem (or, in effect, Searle's criterion of adequacy.)
For Geach the central explanatory notion is that of predication. His crucial claim is "Hence, calling a thing 'P' has to be explained in terms of predicating 'P' of the thing, not the other way round." ('Ascriptivism', p. 253) We have not yet seen his reasons for denying that it can be done the other way round (the ascriptivist way), but it is important here to look at other features of Geach's claim. First, it is true that in every construction containing (some variant on) 'That is good' 'good' functions as a predicate. As Geach is looking for an explanation of the meaning of 'good' that applies equally in every relevant grammatical form, noticing that it is always being predicated of something seems useful.
Before going further it is necessary to issue a warning. Philosophical understanding of what it means to say that something is a predicate typically goes astray. Historically, 'is a predicate' has usually been treated as meaning 'names a property'. So we find for instance, C.J.F. Williams saying (in connection with 'is true') that "our main concern in this chapter is with whether '- is true' can properly be regarded as a predicate."[9] To which the proper answer is 'Of course it is a predicate – look up what a predicate is in any grammar book'. But equally certainly what Williams is doing is conflating, not idiosyncratically, 'predicate' with 'name of a property'. The question he is worrying about is whether 'true' is the name of a property. (His entire chapter illustrates over and over that conflation, even in connection with 'is good'.)
For reasons that will emerge much later, I shall take it that when Geach speaks of predication he must be treating 'predicate' as a purely grammatical notion, not accepting the typical assumption that treats a predicate as (necessarily) a name for a property. And grammatically speaking he is quite correct: 'is good' appears as a predicate in all the grammatical constructions.
Back to his conclusion. What Geach has rejected, for reasons yet to be seen, is the idea that we can explain the meaning of 'good' across all the cases of predicating it by appealing to what is being done when we call a thing good, i.e. he has rejected ascriptivism. What he says must be done in order to achieve a proper account of the meaning of 'good' is to turn that maneuver around, to see that calling the thing good must be explained in terms of the predication of 'good' in all the grammatical constructions. Geach is assuming that a proper explanation of meaning will be (as it were) an equal opportunity employer, that the meaning of 'good' in each and every grammatical form puts no one construction in a privileged position. That, of course, is not what ascriptivism holds: it makes the calling of something good the starting point for an explanation of the meaning of 'good' and all the other cases must (somehow, in a way not yet determined) fall into line. What Geach is looking for is an explanation that treats all the cases equally – and he finds it in the fact that 'is good' is a predicate in all of them.
However, can we really explain what it is to call something good by noting that 'is good' is a predicate? Does that tell us whether 'good' is a property being attributed to the thing or whether ascriptivism is correct in holding that in calling something good a non-descriptive claim is being made about the thing? Certainly not – the choice between those questions cannot be settled by simply noticing that 'good' is a predicate in its uses in various constructions.
Geach is surely partly right: any full (?) philosophical account of the meaning of 'good' must notice that it is essentially a predicate– that we can say 'that is a good peach' or 'the peach is good' but cannot say 'that is a peach good' or 'the good is peach'. So any attempt to explain its meaning by a simple 'It is used to commend' cannot be the full story of our understanding of what goodness is, of how 'good' functions. On the other hand, Geach's replacement explanation captures only part of the truth.
Incidentally, nothing of Geach's attempt to set out the meaning of the 'good' across constructions survives into his second paper on the issues, the paper entitled 'Assertion'. Perhaps he recognized its troubles.
Unlike Geach, Searle does not set out to show how 'good' can mean the same in all the various constructions in which it can occur. However, he tries to do that for one special word and thereby, quite unintentionally, shows one way that ascriptivism might solve the challenge he has made to it.
Searle has a difficulty that Geach did not address (though it is strange that he did not, given the time, place and circumstances of his writing.) Searle is committed to a speech act analysis of 'I promise' and that immediately puts him in the line of fire for the problem that afflicts goodness and other concepts for which ascriptivist accounts have been given. To say 'I promise' is, by meaning, to promise; however, to use that same form of words in any of the other constructions is not to promise ('If I promise, then...' does not make a promise); and yet clearly 'promise' means the same in all those situations. And so the challenge can be offered: how is that possible?
Unlike the case of 'good', Searle needs to defend the possibility of meaning the same across all constructions for 'promise' (and for all the performatives.) How does he show that 'promise' passes the test, that it is possible for 'promise' to mean the same in all the constructions whether or not a promise is being made? His solution: "For example, when one says something of the form, 'If he promises that p, then so and so', one hypothesizes the performance of the act which he performs when he says something of the form, 'I promise that p'." (Speech Acts, p. 138)
That completely fails to do the job of distinguishing between 'promise' and 'good' as he intends it to do. For the ascriptivist can say 'When one says something of the form "If the cake is good, then so and so", one hypothesizes the performance of the act performed when someone says something of the form "That is good".' And so on through the various grammatical constructions.
Does not Searle thereby provide what was missing from the mid-century analyses, namely how to account for the sameness of meaning across all the constructions in which 'good' can occur? His attempt to show how 'promise' means the same across the board has the same starting place as the standard ascriptivist position: that what is done in calling something T is the foundation for understanding the meaning of 'T' where ever it occurs. Contrary to Geach, not all the constructions containing 'T' are of equal importance in setting out the meaning of the term.
On the other hand, Searle's move does not fully help ascriptivism. It merely puts 'promise' and 'good' in the same pot but does not show which pot it is. If one assumes that promising is part of the meaning of 'promise', then Searle's line would show how commending can be included in the meaning of 'good'. However, if one thinks there is good reason for denying that commending is involved in the meaning of 'good', then one will conclude from the assimilation achieved by Searle's account that, perhaps contrary to original impressions, 'promise' cannot, by meaning, make reference to an act of promising. And so the exclusion of pragmatics from semantics will be drawn broadly to include the performatives.
What must be addressed prior to seeing what effect Searle's (unintended) assimilationist move has on ascriptivism are the reasons for holding that the meaning of 'good' cannot involve a reference to commending.
In doing that, the place to begin is by deferring Geach's arguments and looking into Searle first. In attempting to show that speech act analyses of words other than performatives are fallacious, Searle offers the following argument: "But it is equally clear that the speech act analysis of the other words; 'good', 'true', ...etc. does not satisfy this condition [of adequacy]. Consider the following examples: 'if this is good, then we ought to buy it', is not equivalent to 'If I commend this, then we ought to buy it'. 'This used to be good' is not equivalent to 'I used to commend this'. "I wonder whether this is good' is not equivalent to 'I wonder whether I commend this', etc. Similar counter-examples will refute the [other] speech act analyses." (Speech Acts, pp. 138-139)
Of course, Searle is right about those non-equivalences. However, is his representation of what ascriptivism amounts to correct? Not at all: not even in the case where something is being called good. To produce his objection Searle converts 'That is good' wherever it occurs into 'I commend that'. Though his immediate targets, the mid-century ascriptivists, were confused about the relation between performatives and speech acts generally, it doesn't follow that the ascriptivist pattern itself must be confused about that. For the commendation performed by calling something good is not to be represented by paraphrasing 'That is good' as 'I commend it'.
Calling something good stands in contrast to talking about one's own response to the thing. The point of saying 'That is good' is to get outside of oneself, to direct attention to the thing and its qualities rather than to oneself and one's response to it. Both Searle and the mid-century emotivists (Ayer, Stevenson) fail to notice that contrast when they assimilate 'That is good' to 'I commend it' or to 'I like it'. That failure shows up explicitly when Searle tries to paraphrase one of the other constructions, 'I wonder if that is good', as 'I wonder whether I commend that'. Searle takes that lack of equivalence to show that calling something good cannot involve commending. However, it should be taken to show that he misunderstands how the act of commending is related to the use of the word 'good'.
For the ascriptivist the commendation is built into the act of calling something good and the aim in calling something good is not to talk about one's self but rather to be referring to the thing. To wonder whether something is good is to be wondering whether the thing is commendable and not to be considering whether I have commended it.
That realization supplies the clue as to how we should regard the disputed relation between the meaning of 'good' in 'That is good' and its meaning in its other possible occurrences, how the ascriptivist is to reply to the original challenge. When we say 'If that is good, then...' we are saying 'If that is commendable, then...' Similarly, for the other constructions (e.g. 'That used to be good' means 'That used to be commendable'). It is possible to have 'good' mean the same in 'That is good' and 'If that is good' while being a commendation in one and not in the other: it is the notion of commendability that binds them together.
It is at precisely this point that Geach's criticism of holding that commendation is central to the meaning of 'good' becomes relevant. It is Geach's criticism that is most heavily relied upon by current critics of ascriptivism. For Geach, to accept ascriptivism about 'good' would require that the term mean something different in other constructions than it does in calling something good. "That would mean that arguments of the pattern 'if x is true (if w is bad), then p; but x is true (w is bad); ergo p' contained a fallacy of equivocation, whereas they are in fact clearly valid." ('Ascriptivism', p. 253)
I committed myself, and ascriptivism, to the following claim: contrary to Geach, the argument form 'if that is good (i.e commendable), then p; but that is good (a commendation); ergo p' is valid, does not contain a fallacy of equivocation. That needs defense.
Alexander Miller in utilizing Geach's objection holds that the ascriptivist position on the matter is comparable to 'My beer has a head on it; if something has a head on it, it must have eyes and ears; hence my beer has eyes and ears'.[10]
Now that is certainly a case of equivocation, the word 'head' being used in two different senses. But it is not an analog of the ascriptivist argument. Miller's supposedly parallel argument does satisfy the standard definition of equivocation: there is one word, 'head', with two different meanings. One of the meanings was presumably derived from the other analogically – but that it became an established usage is a historical accident, a relation not required for understanding talk of a beer's having or not having a head. On the other hand, while there is one word in the supposedly comparable case concerning 'good', its meanings in the two premises are essentially (internally) related: commending and being commendable are part of a single family, they are grammatical variants of the same concept.
The relevant comparison is more like this argument: 'if the pipe breaks, the room will be flooded; the pipe broke; hence the room is flooded'. Or 'If the pipe breaks, the room will be flooded; the pipe is breaking; hence the room is being flooded'. Or 'if the pipe were to break, the room would be flooded; it did break; hence the room is flooded'. All of those arguments, despite the small changes in wording, are valid – because all the terms are part of the same families, exhibiting differences only in tenses.
C.J.F. Williams, speaking for logic, puts it this way. "The last example shows that we [logicians] need not be pernickety about what counts as a repetition of the same sentence.... Sometimes repetition of exactly the same words is idiomatically impossible or clumsy in English.... English idiom does not allow us to say 'Percy said that Charles would grow out of his shyness and Charles would grow out of his shyness." Again: "But logicians...are as much entitled to take 'Percy said that Charles would grow out of his shyness and Charles has grown out of his shyness' as a substitution instance of 'Percy said that p and p' as they are to take 'She thought that Barbara ought to pay her' as a substitution instance of 'x thought that Barbara ought to pay x' without boggling at the change of case from 'she' to 'her'."[11]  
Only if one is "pernickety" about sameness, only if one has extremely narrow ideas of what must be kept the same in argumentation, will one be inclined to regard the change from 'good' as commending to 'good' as commendable as a case of equivocation and thus the creation of an invalid argument. Williams is right, logic is entitled to treat those arguments as valid – because they are valid and consequently their formal representation must treat them as valid. Logicians, just as the rest of us, are entitled to reject Geach's contention and accept the variation in terminology as legitimate when the terms are internally related, i.e. related in meaning, versions of the same concept.
The upshot is that Geach's criticism of the ascriptivist position about 'good' in particular and about the pattern generally is misguided. And since it is that criticism that has seemed to be so powerful, to deal a death blow to any attempt to indulge in ascriptivist analyses, it follows that there is no case to be made from the Geach-Searle perspective against ascriptivism.
While the standard criticisms of the ascriptivist contention - that a reference to commending is a central feature of the meaning of 'good' - do not succeed, nonetheless the ascriptivist view is not the default position and does not automatically succeed when criticism of it fails. More needs to be said in both explanation and defense of ascriptivism.
The mid-century analysts did not attempt what needs to be done. For them it was a matter of insight that to understand 'good' requires grasping its connection to commending. While insight in philosophy is not to be sneezed at (although, with our heavy emphasis on argumentation, it typically is), nonetheless surely something can be said in favor of the claim over and above 'The criticism of it fails' and 'Don't you see?'.
In providing a case for that pattern of philosophizing, especially in its application to goodness, it is necessary to begin with a topic that is ignored by critics. One piece of the background to the origin of ascriptivism in the mid-20th-century is, as has already been mentioned, Austin's recognition of the special kind of function 'I promise' and other performatives have. Equally, however, Moore's denial that goodness is a natural property of things played a crucial role in the development of ascriptivism. As I presented the ascriptivist position earlier, its starting place is the denial that goodness is a property (though of course 'good' is a predicate), the denial that 'good' names, refers to, a feature, characteristic, attribute of things. That thesis, which originated as a response to Moore, is an important element in any version of ascriptivism, including the one being defended here. For it creates the presumption that 'good' performs some function other than that of designating or naming.
It has long been recognized that Moore's self-named naturalistic fallacy is mis-named. Others have pointed out that he should have called it the definist fallacy. He was denying that it is possible to define 'good'. Any such a definition is intended to be reductionist: the word 'good' would be held to mean the same as a word/phrase where the defining term is the name of, stands for a property. Goodness, on that view, is identical with that property, is nothing other than, over and above, that property. Ascriptivists, then and now, accept the Moorean thesis that 'good' cannot be defined, i.e. that it is not equivalent in meaning to any term designating a property, even in the very broad sense of 'property' intended, and that consequently goodness is not a property of things.
For Moore, however, to hold that 'good' cannot be defined was not the same as claiming that goodness is not a property. Though satisfied with his criticism of the definist position, he continued to hold that good is a property, albeit a non-natural property of things, a simple and undefinable property comparable to yellow. It is at this point that ascriptivism digs in its heels and departs from Moore. He implicitly accepts the traditional assumption that a predicate names a property and 'good', being a predicate, must therefore be a designator. While we can examine something and discern via our senses its yellowness, no matter how thoroughly we inspect a thing that we are agree is good, we cannot similarly locate its simple property of goodness. Since goodness is a property and is not identical with a property bearing some other designation ("Every thing is what it is and not another thing") and yet is not observable, it must be a non-natural property, something accessible only by a faculty of intuition.[12]  
The ascriptivist rebels here. Ascriptivism is a naturalistic position: rather than resorting to non-natural transcendent properties discernible only by a seemingly invented faculty, it is simpler and more cogent to surrender the idea that goodness is a property and thus treat the objections to defining it as establishing that conclusion. Of course, that does leave the problem of what status the predicate 'good' has – and ascriptivism is only too happy to supply a radically non-traditional account.
So the outcome hinges on objections to defining 'good'. If the definist project, and not just particular definitions, fails, then the idea that goodness is a property fails. Moore, of course, came to the conclusion that 'good' cannot be defined by employing the open-question argument: any attempt to define 'good' fails because it will always make sense to ask 'It has the property P but is it good?' It is today widely accepted that the open-question argument does not do what Moore assumed it did, namely provide a complete refutation of the idea that 'good' can be defined. Instead, it is merely the beginning of an extended and ramified discussion between supporters and opponents of the idea that 'good' can be defined.
Since I have claimed that an up-to-date ascriptivism about 'good' requires that goodness not be a property and that the possibility of defining it requires that it be a property, that is, that goodness be identical with a property whose basic designation is something other than 'good', ascriptivism must defend the Moorean conclusion. On the other hand, obviously, given the direction of philosophical thought since Moore one cannot accept the open-question argument as constituting the only line of defense of that conclusion. That it does seemingly always make sense to wonder whether the supposed defining property and goodness are the same, whether 'P' and 'good' mean the same, the open question argument can be no more than one item in support of Moore's conclusion.
I shall consequently produce a larger set of considerations, set out below in no particular order and having different degrees of significance, that ascriptivism today would rely upon to show that 'good' cannot be defined and that therefore goodness cannot be a property.[13]  
Start here. Although it is a clever move to have one's favored claim about some subject matter smuggled in as a matter of meaning, one must be suspicious about such definitions. Why should not the 'definitions' of 'good' be regarded as substantial claims about what makes something, anything, good rather than as explications of the term? There seems to be nothing in the definist literature addressing that question. It appears that the aim of the definist program, especially in its resurrected naturalistic (scientistic) form, is to get rid of, by reduction, a notion not amenable to scientific investigation.
It is often said that goodness is a moral concept. "Moore sets out to argue that moral terms such as 'good' are not definable...."[14]  But 'good' is not a specifically moral term - it is applicable to items of many other kinds than to matters of morality. Speaking of a good cigar or a good essay is as much relevant to the question of what 'That is good' means, what its function is, as its reference to a person, action, etc. ('He is a good father', 'That was a good deed'). In fact, given the recognition by Oliver Wendell Holmes that hard cases make bad law, we can say that hard cases, matters of morality, make bad theses about meaning. To investigate the meaning of 'good' we should chiefly concern ourselves with what it means to say, e.g. 'That is a good peach'. However, the definitional enterprise does not do that. The great gap between the sorts of answers that that enterprise results in, whether naturalistic or theological, and what is going on when we consider whether a given peach is good produces deep suspicions about the legitimacy of the definist project.
As is widely recognized, what turns up when we think about what counts as a good peach are its properties (flavor, texture, etc.). But when we also consider what counts, for example, as a good shortstop and what makes for a good philosophical essay, we quickly realize that there is no overlap in the properties cited in all the various cases. Since none of those good-making properties can plausibly be taken as contained in the meaning of the word 'good' - it is applicable to so many different kinds of thing - but are rather what is mentioned in the differing judgments of goodness about different kinds of things, to cigars and essays and shortstops, it certainly seems that goodness cannot be understood in terms of specific defining properties.
Recalling that to talk of (say) a good philosophical essay in ordinary practice makes reference to characteristics of the essay, a further difficulty for the definist enterprise comes out. How do those various attempted definitions of 'good' relate to the properties cited in our ordinary judgments of the goodness of a cake or a piece of philosophy? Are they being recommended as replacements? Are they what 'good' really means contrary to our practice? Are they generalizations of what we are talking about when we cite the juiciness of the peach and the argumentation of the philosophical essay? Is it simply not recognized by the definist that we do make reference to such properties when we call things good? Once again, the status of the results of definist projects are not spelled out in the literature.
While Geach's objections to the ascriptivist pattern are vital to today's definist tradition, the definist overlooks his contributions to the discussion in his influential 'Good and Evil'. Although the main target there is again ascriptivism, his arguments entail a powerful criticism of attempts to give a definition of 'good'.
Geach notices that from 'That is a yellow rose' we can infer 'That is yellow' but that from 'That is a large flea' we cannot infer 'That is large'. He holds that 'good' follows the pattern of 'large' and not of 'yellow': from 'He is a good assassin' we cannot infer 'He is good' simpliciter. It does not follow, of course, that 'good' is not a property word, for then we would be committed to 'large' not being a property word either: and it clearly is. But what we can conclude is that goodness is relative to the kind of thing being characterized as good. That means that there is no such property as goodness independently of the applicability of 'good' to various kinds of things. As we have seen, what makes this kind of thing good is not what makes that kind of thing good. There is no property generally definable as constituting goodness. There is no Form of the Good.
There is more to say about Geach's argument in 'Good and Evil' but further examination of that becomes relevant later. For now let me offer one more line of criticism that an up-to-date ascriptivism would apply to the idea that goodness is a property whose character we can spell out in a definition. Ironically, this line of criticism is derived from the very philosophers who attempted to refute ascriptivism, namely Geach and Searle. Possibly they themselves saw the difficulty and avoided it in their own attempts to say how 'good' can mean the same across the constructions in which it can occur. But the referentialists that Moore was responding to and the referentialists who have resurfaced following the rejection of ascriptivism by Geach and Searle all fall into the trap.
The problem set by Geach and Searle was that if 'That is good' is commendatory in its meaning, those same words are not a commendation at all in, say, 'If that is good': yet 'good' means the same in both places. Now suppose that, on the contrary, we agree with the tradition that the function of calling something good is to attribute a property to it. However, when we use the same words with the same meaning to say 'If that is good' those words there are not attributing a specific property to the thing, are not describing it (a point that is an echo of Frege on what the role of predication). In other words, the resurrected traditional view that 'good' names a property falls victim to the same criticism that Geach and Searle level against ascriptivism. Giving up on ascriptivism does not enable one to assume, as is typically done, that it is thereby established that goodness is a property, is identical with some property.
Of course, those treating good as a property can take up the same challenge as Geach-Searle present to the ascriptivist: how is it possible to have the word mean the same yet in one case it attributes a property to a thing and in other constructions not? I say they can take up that challenge – they have not done so, have not recognized the difficulty. They might make use of Searle's tactic of saying that in the conditional the same act (of attributing a property) is hypothesized as was performed in calling the thing good. Consequently, the meaning can be seen to be the same. However, it is not certain that a philosopher currently in the definist tradition wants to join Searle (and the ascriptivist) in making calling something good primary in meaning among all the grammatical constructions and linguistic activities and thereby making an act, a deed, part of the meaning of the term 'good'.
Since the definist has not addressed the issue I cannot predict what they might do by way of response. However, the current ascriptivist position is that, until something is produced by way of response to the challenge, something that is worthy of serious thought, then ascriptivism holds the field. For that reason and for all the other considerations offered above, we should reject the idea that 'good' refers to a property and can be defined in terms of that property.
There is another possible wrinkle in the definist research program. Recently, perhaps from the anxiety that the open-question argument cannot be completely dismissed, that the Moorean insistence that 'good' does not mean 'P' is nothing but conceptual confusion, some of today's definers have attempted to employ a different model of definition. Rather than holding that 'good' means "P', they have resorted to Frege's Sinn and Bedeutung distinction. On that view, 'good' and 'P' do not mean the same – rather 'good' and 'P' have different senses but they refer to (name, stand for) the same property.[15] 
That 'good' is referential, is a property word, is simply assumed there as it was when the other version of definition was employed. The battery of arguments above to show that it is not the name of a property are not recognized. In consequence, the resort to the Fregean model will not do to save the definist position.
Moreover, because the good-making characteristics of a kind of thing are tied to the kind of thing it is, and are not a character applicable across the board which is attached to the word 'good', are not part of the meaning of 'good', there is no single mode of presentation for 'good', no Sinn for 'good'. Hence 'good' is a word with neither a Sinn nor a Bedeutung.
Having now disposed of the traditional alternative to the thesis that commending is a central feature of the meaning of 'good', the time has come to take up the task of defending that thesis. I am simply ignoring another alternative, one that was also the off-spring of Moore's anti-definist views, namely emotivism. The objections to that position are too powerful to take it seriously at this date.
Let us return to Geach's 'Good and Evil'. The non-validity of the inference from 'She is a good player' to 'She is good' reminds us that what makes some kind of thing good varies radically from kind to kind and that there are no good-making characteristics that are applicable to everything that is good. Geach is tempted by that fact to say that 'good' has no meaning other than in 'a good K'. But of course that isn't so. If it were, there would be no intelligibility in the fact that we use the same word 'good' in connection with oranges, designs, helicopters and on and on. Moreover, the word 'good' would be enormously equivocal – and it is not.
Geach, although tempted to talk as if 'good' does not mean anything independently of its connection to kinds, realizes that won't work, no doubt for the above reasons. But if 'good' means the same across kinds of thing, the problem of what that common meaning is rears its head. It would appear that here is where the ascriptivist idea that commending is central to the meaning of the term finds a place.
Geach, however, will have none of that: even though he doesn't yet have in his pocket the objections developed in 'Ascriptivism' to recognizing commendability to be involved in the meaning of 'good'. So he needs a different solution to the problem of what 'good' means in its applicability to hugely different kinds of thing. He finds a solution that enables him both to show how 'good' can be descriptive without making it ambiguous (equivocal) without including commendability as the, or even a, common feature of the meaning of 'good'. His solution is to hold that 'good' functions like 'is the square of'. ('Good and Evil', p. 69) Robert Arrington spells out how this is meant: "But what they have in common need not be an identical and independent evaluative meaning. According to Geach, 'good' could have a common abstract descriptive meaning, but one that gets fleshed out only in relation to the particular kind of thing that is called 'good'. It could be a common function like the squaring function, that gets its concrete meaning only in relation to its specific values."[16] 
There are two important points to be made in connection with Geach's solution. First, while the analogy of 'good' to 'is the square of' is very useful, more must to be done to make it applicable to the case of 'good'. In order to explain how 'good' can be applied with the same meaning to all manner of things possessing widely different good-making characteristics, mentions of which are not included in the meaning of the term, one must say something like this: the word 'good' has as a central feature of its meaning the requirement that to call a particular thing good must be done in light the things possession of certain properties, with the further understanding that the relevant characteristics will vary from kind to kind. Those who have mastered the concept of good will realize not only that certain features of a thing need to be cited to justify a claim that it is good but also that those features vary widely depending upon the kind of thing in question. And we surely do know that: we who know the meaning of 'good' know both that some features of a thing must be cited to justify the ascription of goodness to that thing and also that, for example, what makes a good chocolate cake is not what makes a good joke.
But there is a second point to be made. Why must we choose between Geach's solution to the meaning of 'good' across kinds of thing, which is surely correct, and the idea that what makes it intelligible that so many different kinds of thing are good for so many different kinds of reason is that when we call a thing good we are commending it? There is nothing in comparing 'good' to the squaring function nor in the more detailed extension that I have given above that requires that Geach's solution to the issue of the common meaning of 'good' is in opposition to the ascriptivist view. Only the objections later developed to the inclusion of commendability in the meaning of 'good' stand in the way of including both considerations in the meaning of 'good'. Since those objections fail, as we have seen, there is nothing standing in the way of saying that that the concept of goodness, the meaning of 'good', is to be understood as the commendability of a thing in light of its good-making characteristics which vary radically from kind to kind.
Notice: in that account of the meaning of 'good', there is no distinction between two different kinds of meaning as there was in the early version of ascriptivism. The distinction between evaluative meaning and descriptive meaning vanishes, although involved in the meaning of 'good' are both a reference to commendability and a requirement that features of the kind of thing in question must be cited in justifying a claim to its goodness. The old disputes about two kinds of meaning and which has precedence simply disappear.
That specification of what 'good' means satisfies several concerns. The Geach-Searle criticism of putting commendation into the meaning has failed and that is reflected in the above account. The need to include reasons for calling something good is made a central feature of the meaning of the term. We can thereby explain how it is that good has a single meaning and is not wildly equivocal. Understanding the meaning of 'good' in that manner works with the fact that 'good' can occur in a many different grammatical constructions ('If that is good', where of course 'That' refers to some particular kind of thing, i.e. some X, means roughly 'If it is commendable in light of its possession of good-making features as built into the meaning of 'X', then....').
Recall that none of the parties to the dispute deny that to call something good is to commend it. But the mid-century analysts saw that that was built into the very meaning of 'good' but did not see that they had to continue from that point. Geach thinks that we can say that calling a thing good is commending it but denies that that is built into the meaning of 'good' and that the fact of commending is (thus) philosophically irrelevant. Searle sees that commending is not externally connected to calling something good but offers no help, at least not intentional help, in understanding how to proceed from that point. The semantics-pragmatics distinction, erected in light of Geach, holds that the commending falls outside the meaning and belongs to a different realm of language (use) and so may or may not be of philosophical interest.
We talk of 'good-making characteristics' of a kind of thing. But what sense does it make to mark out certain properties as having that kind of interest so that we can put them in a special category, as the properties that make a thing good? Only by seeing that we mark them out thusly in order to commend things of that kind that have those features can we make sense of the practice of creating the category. That is, there is no sense to talk of being good unless we can explain why those features are special. And the definist, who wants to eliminate the notion of goodness by reducing it to some property, natural or theological, thereby does not have to explain why we have the category of goodness, why we talk and think in terms of goodness instead of just saying the thing has a property P. Commendability must be included in the concept of goodness in order to explain the existence of the concept.
The line between semantics and pragmatics, where it is taken that a reference to deeds cannot be part of the meaning of a term, hangs on the Geach-Searle criticism. Seeing that that criticism does not succeed means that the sharp line between meaning and what we do with words cannot be maintained. Although some aspects of the use of a word may not fall into its meaning, some references to what we are doing must be located there. As Wittgenstein was fond of saying in light of Faust 'In the beginning was the deed'.



1. The relevant writings are: Peter Geach, 'Ascriptivism', Logic Matters (Blackwell, 1972), pp. 250-254, originally published in Philosophical Review 69 (1960), pp. 221-225; Geach, 'Assertion', Logic Matters, pp. 254-269, originally published in Philosophical Review 74 (1965), pp. 449-465; John Searle, Speech Acts, (Cambridge University Press, 1969), pp. 136-141; (see p. 70 of Speech Acts for a reference to 'Ascriptivism'.) A related earlier paper, largely overlooked in these discussions, is Geach, 'Good and Evil', Theories of Ethics (Oxford Readings in Philosophy), ed. Philippa Foot, (Oxford University Press, 1967), pp. 64–73, originally published in Analysis 17 (1956), pp. 33-42.  To save time and space I shall refer to page numbers of those works in the body of the text, citing item and page, e.g. ('Ascriptivism', p. 254.)

2. Alexander Miller, An Introduction to Contemporary Metaethics, (Polity Press, 2003).

3. It will be noticed that I speak of the Geach-Searle objections. Miller calls the objections and the view they express the Frege-Geach problem. Geach does claim Frege as his predecessor. Frege, in talking of predication, holds that the traditional definition of a predicate as that which we use to say something of a subject is mistaken. His reason is that the same predicate with the same meaning can occur in the antecedent of a hypothetical and there it is not being used to say something of a subject – in 'If Frege wrote best-sellers, then...' we are not saying anything of Frege unlike in 'Frege wrote best-sellers'. Geach finds in Frege materials for his attack on ascriptivism. But it is Geach's – and Searle's – attack on a position that was created only in the 1940's and 1950's that is so potent today and that has led to the sharp distinction between semantics and pragmatics. Hence it is best to keep Frege in the footnotes as an ancestor and think of the issues as those defined by Geach and Searle. Incidentally, Miller curiously makes no reference to Searle, not even adding a mention of Searle as I do here of Frege.

4. H.L.A. Hart, 'The Ascription of Responsibility and Rights', Proceedings of the Aristotelian Society New Series, Vol 49 (1948-49), pp. 171-194; reprinted in Anthony Flew ed., Logic and Language, 1st Series, (Blackwell, 1951).

5. C.J.F. Williams, What is Truth?, (Cambridge University Press, 1976, reprinted 2009), p.16.

6. Beyond not setting out the pattern of thinking, neither Geach nor Searle give a satisfactory specification of which particular accounts of concepts exhibit that pattern. Geach is especially interested in Hart's views – Searle does not mention Hart. They agree that Strawson's account of 'true' and Hare's account of 'good' belong to the pattern. Searle adds to his list Austin on 'know' and Toulmin on 'probably'. What is omitted is also important. There is no reference to Austin on 'I promise', though that surely belongs to the pattern (although Searle, as we shall see, makes an effort to deny that.) There is no mention of Wittgenstein where his discussions of 'pain' were a staple of the times and clearly part of the same pattern of analysis, as was the discussion of the builder's language at the very beginning of the Philosophical Investigations. What binds all those discussions into a pattern of philosophizing? There is a negative aspect: the core cases involve a denial that the relevant words are names, that the words function to refer to some property of things. On the positive side, the various analyses are in agreement that the relevant terms have as a central feature of their meaning the performance of an act, a deed.

7. Ryle introduced the notion of logical geography as a replacement notion for the older idea that analysis was like chemical analysis, a breaking down of a substance into its ultimate parts. Gilbert Ryle, The Concept of Mind, (Hutchinson's University Library, 1949), p. 7. Strawson talks of connective analysis as a better way of conceiving of the activity of analysis than the old reductive version or Ryle's logical geography or Wittgenstein's therapeutic analogy. Peter Strawson, Analysis and Metaphysics, (Oxford University Press, 1992), Chapter 2.

8. My claim that judgments of the form 'That is good' have a truth-value goes beyond the argument presented above. For that argument is only that calling something good cannot be held to lack a truth-value on speech-act grounds. While that is the usual ground presented for holding that the ascriptivist view is non-cognitive, it is easily possible to find others. The chief one will be that ascriptivism, as mentioned previously and argued later, denies that goodness is a property, either under that name or some other. It will be said that if not a property, then without truth-value. Such a claim assumes the correspondence theory of truth as an account of what is being done when we can something true or false, that we are claiming a correspondence between what is said and a fact. Now that is, at least, eminently discussable. Nothing, however, in this paper hinges on that matter. By my saying that ascriptivism is a cognitive view I am signaling that I would offer a different account than the correspondence theory of what is going on when we talk of truth or falsity. For present purposes all that is necessary is to realize that I refuse to define ascriptivism in terms of being non-cognitive.

9. Williams, op. cit. p. 29.

10. Miller, op. cit., p. 41.

11. Williams, op. cit., pp, 2-3 and 4.

12. In treating intuition as a faculty, I am taking sides in a dispute concerning how to conceive of what Moore meant by 'intuition'. Pursuing the dispute here would be too large a side-show.

13. Miller does not consider any other objections to giving a definition of 'good' than the open question argument. The additional objections that I offer are not recognized and so not evaluated.

14. Miller, op. cit., p.12.

15. Miller, op. cit., pp. 17-18.

16. Robert Arrington, Rationalism, Realism, and Relativism: Perspectives in Contemporary Moral Epistemology, (Cornell University Press, 1989), p. 56.