Jeffrey C. King

2004 Northwest Philosophy Conference

A. Questions any account of analysis must answer to be a contender:

i. What are the objects of analysis (that is, what is analyzed in a correct philosophical analysis)?

ii. Under what conditions is a purported analysis a correct analysis?

iii. Why are we able to produce philosophical analyses from the armchair? (I am deliberately staying away from the term ‘a priori’ here—more on this below.)

iv. What is the difference between the following sorts of claims, all of which seem to deserve the label ‘analysis’ but each of which differs from the others in certain ways:

a. For all x, x is a brother iff x is a male sibling.

b. For all x, x is knowledge iff x is a justified true belief.

c. For all x, x is water iff x is liquid H₂O.

v. What is the solution to the paradox of analysis?

B. Further comments on questions iv and v: regarding iv, though a-c all seem to be in some sense analyses, things like a are trivial in a way things like b and c aren’t. Things like b are discoverable from the armchair (like a and not like b). Things like c require empirical investigation (unlike a and b). As to question v, various things have been called the paradox of analysis over the years, but I think there is a common structure. We begin with something that is claimed to be a correct analysis, say:

(1) For all x, x is an instance of knowledge iff x is a justified true belief.

(2) For all x, x is a brother iff x is a male sibling

It is then claimed that if (1) and (2) are correct analyses, we may infer that they must "mean the same thing as" or "say the same thing as" or "express the same proposition as" or "make the same statement as"

(1a) For all x, x is an instance of knowledge iff x is an instance of knowledge

(2a) For all x, x is a brother iff x is a brother

It is then claimed that there is some difference between (1) and (1a), (e.g. one is informative, the other not) and (2) and (2a) (e.g. one is an analysis, the other not) that precludes their "meaning the same thing", etc. The paradox is that given that (1) and (2) are correct analyses, it appears that the sentence pairs (1)/(1a) and (2)/(2a) must and must not "mean the same thing", "express the same proposition", etc.

A. The framework to be employed (three elements)

1. The theory of propositions

a. I will assume the propositions are structured entities, and that the structure of a proposition is at least very similar to the structure of the sentence expressing it. In fact, on the view of propositions I have defended elsewhere, the structures of propositions are identical to the syntactic structures of sentences expressing them (at the relevant level of syntax). Indeed, on my view something even stronger can be said. But I don’t want my view on analysis to depend on my theory of structured propositions, so that someone who rejects my view of propositions may endorse my account of analysis. Thus, I here assume only that the structures of propositions are similar to the structures of sentences that express them.

2. Some properties and relations are complex, and have other properties and relations as components.

a. To illustrate, take the property of being a bachelor. It might be thought that the properties of being adult, unmarried, human and male are components of the bachelor property. The idea is that the bachelor property is a complex property that has being unmarried, being adult and so on as component parts that are combined conjunctively to form the bachelor property.

b. In this case, the mode of combination of the component properties that form the bachelor property is something like property conjunction. But it should not be thought that conjunction is the only mode of combination. For, first, I at least would countenance a disjunctive mode of combination. But more importantly, there are other modes as well. Consider the property of being an uncle. Roughly, we might think the uncle property is to be understood as (suppressing being human for simplicity):

3. x is an uncle iff x is male and some:y( x is a sibling of y & some: z( y is a parent of z))

Here there are various modes of combination that combine the components in forming the uncle property. First, the parent of relation combines with the property of properties some. Some combines with a two place relation on one of its argument places to yield a one place property. In the above case, some combines with the parent of relation on its second argument place to yield a one place property that holds of x iff x is a parent of something. Similar remarks apply to some combining with the two place relation of x’s being y’s sibling where y is a parent.

3. There are (at least) three categories of words governed by different standards of linguistic competence.

a. There are three categories of words such that the words in a given category are all governed by the same standard of linguistic competence; but words in different categories are governed by different standards of linguistic competence.

i. Category 1: to be competent with a word in the first category, which I will call a category one word, one must be able to specify the components of the property or relation expressed by the word and how those components are combined in the property expressed by the word. ‘Bachelor’ is a paradigmatic category one word. To be competent with the word, one must know that it expresses a property that has as components the properties of being unmarried, being male and being adult and that these properties are "conjunctively combined" in the bachelor property.

ii. Category 2: very roughly, I want to say that to be competent with a category two word w requires that one be able to reliably determine whether a given entity possesses the property expressed by w and to thereby know whether w applies to the entity or not. More explicitly, being competent with such a word w requires that given any entity o that paradigmatically possesses the property P expressed by w or paradigmatically fail to possess P, were one to know all facts about o relevant to whether it possesses P or not, excluding the fact that o possesses P or the fact that o fails to possess P, one would correctly determine whether o possesses P and w applies to it or not. Let’s stipulate that category two words can’t be in category one as well. So if something is a category one word it ipso facto fails to be in category two.

iii. Category 3: words in category three fail to be governed by the standards of competence governing categories one and two. Indeed, I think it is hard to say what the standard of competence is for such words. Let me just say that the paradigmatic category three words are the so-called natural kind terms like ‘water’, ‘tiger’, aluminum’ and so on.

B. Some additional matters

1. I shall assume that analyses are stated in the following canonical form:

4. For all x, x is F iff C(x)

where I shall assume that ‘F’ is syntactically simple and that ‘C(x)’ is a syntactically complex predicate containing free occurrences of the variable ‘x’. Now according to our theory of propositions, sentences like 4 express propositions whose structures are very similar to the structures of the sentences expressing them. So 4 expresses a proposition that looks something like this:

4’. [EVERY: x [F’(x) IFF C(x)]]

where EVERY is the contribution ‘For all’ makes to propositions, IFF is the contribution ‘iff’ makes, and F’ is the property expressed by ‘F’. I’ll come back to these constituents in a moment. First, let me discuss C(x). C(x) is what ‘C(x)’ contributes to 4’. As I indicated above, ‘C(x)’ is syntactically complex. Since we have assumed that the structures of propositions are very similar to the syntactic structures of sentences expressing them, the structure of 4’ must be very similar to the structure of 4. In turn, this means that the structure of C(x) must be very similar to the syntactic structure of ‘C(x)’. I take C(x) to be a structured complex propositional constituent whose structure mirrors the syntactic structure of ‘C(x)’. In general, syntactically complex predicates contribute to propositions complex propositional constituents whose structures mirror the syntactic structures of the predicates contributing them. Let me illustrate this with a relatively simple example. Consider the complex predicate:

5. met Seth Morrison

Oversimplifying, this predicate has the syntactic structure as follows:

5’.

met Seth Morrison

Now on our view of propositions, the syntactic structure of a sentence like

6. Every skier met Seth Morrison

is at least very similar to the structure of the proposition expressed by that sentence. But this means that the contribution made by the predicate 5/5’ to the proposition expressed by 6 must have a structure at least similar to 5’. My own view is that the contribution made by the predicate 5/5’ to the proposition expressed by the sentence 6 is the following entity:

7.

met* Seth Morrison*

where met* is the relation of meeting and Seth Morrison* is Seth Morrison. Let’s call the things at the terminal nodes of entities like 7 its constituents. So 7s constituents are met* and Seth. Clearly, the entity 7 has the same structure as the predicate 5/5’ as required. Now 7 is not the relational property of meeting Seth Morrison. Presumably, that relational property is the result of Seth saturating the second argument place in the meeting relation. That is not what 7 is. 7 is the meeting relation standing in a relation, represented by the bracketing, to Seth Morrison. The relation that holds together Seth and the meeting relation in 7 (again, represented by the bracketing) is of the same sort that generally holds together the constituents of propositions.

Though 7 is not the relational property of meeting Seth, in the definition of truth for propositions, the sub-propositional constituent 7 must get mapped to that relational property. After all, ‘Every’ in a sentence like 6 contributes to the proposition expressed by it a relation between properties. Thus, 6 expresses a proposition that is true iff the properties of being a skier and the relational property of meeting Seth stand in the relation expressed by ‘Every’. So in the definition of truth for propositions, the sub-propositional constituent 7 must get mapped to the relational property of meeting Seth. And generally, complex predicates will contribute entities like 7 (except more complex if the predicate is syntactically more complex than 5/5’) to propositions, and the definition of truth for propositions must map these entities to the relevant properties. I shall put this by saying that the complex sub-propositional contributions syntactically complex predicates make to propositions, though not properties or relations themselves, represent properties or relations.

Returning to C(x), what I have just been saying is true of it, since it is the contribution made to the proposition 4’ by the syntactically complex predicate ‘C(x)’ in 4. Thus, C(x) is a complex sub-propositional entity like 7 except perhaps more complex; and like 7, it represents a property, that is, gets mapped to the property by the definition of truth for propositions.

2. Other issues regarding 4’: I’ll assume we have an account of determiners and sentential connectives according to which sentences of the form of 4, and hence propositions of the form 4’, are true iff everything is such that it possesses F’ iff it possess the property represented by C(x). Obviously, then, many true propositions of the form of 4’ fail to be analyses.

C. The account of analysis

1. An analysis is a proposition of the form 4’ such that (i) the property represented by C(x) is identical to the property F’; and (ii) the constituents of C(x) are components of the complex property F’.

a. On the view in question, what is being analyzed is a complex property or relation. In the case of 4’, it is F’. The constituents of the entity C(x) are components of the complex property F’ being analyzed. So the analysis tells us what the components are of the complex property being analyzed. If we assume, as seems very plausible, that a complex property or relation has the components it does as a matter of necessity, then an analysis will be a necessarily true proposition. This answers questions i and ii we began with.

b. Turning to question v and the paradox of analysis, recall that in order to get the paradox, we need the claim that if the following are analyses

(1) For all x, x is an instance of knowledge iff x is a justified true belief.

(2) For all x, x is a brother iff x is a male sibling

then they must "mean the same thing as" or "say the same thing as" or "express the same proposition as" or "make the same statement as"

(1a) For all x, x is an instance of knowledge iff x is an instance of knowledge.

(2a) For all x, x is a brother iff x is a brother.

But we reject this claim, because on the present view 1 and 1a (and 2 and 2a) do not express the same proposition even if 1 (and 2) is an analysis. The reason is that ‘knowledge’ contributes the property of being a bit of knowledge to the proposition expressed by 1, whereas ‘justified true belief’ contributes a complex C(x) type entity that represents the property of being a bit of knowledge (assuming 1 is an analysis). Thus, on the one hand, ‘knowledge’ and ‘justified true belief’ make different contributions to the proposition expressed by 1, and so we block the paradox of analysis by holding that 1 and 1a don’t express the same proposition. But on the other hand, since ‘justified true belief’ contributes to the proposition expressed by 1 a C(x) type of entity that represents the property of being a bit of knowledge, we preserve the intuition that ‘knowledge’ and ‘justified true belief’ in some sense "stand for" the same property.

D. The account of philosophical analysis

1. This leaves only our original questions iii and iv unanswered:

iii. Why are we able to produce philosophical analyses from the armchair?

iv. What is the difference between the following sorts of claims, all of which seem to deserve the label ‘analysis’ but each of which differs from the others in certain ways:

a. For all x, x is a brother iff x is a male sibling.

b. For all x, x is knowledge iff x is a justified true belief.

c. For all x, x is water iff x is liquid H₂O.

a. First, consider c in question iv. If water is indeed liquid H₂O, then this seems like an analysis in some sense. And perhaps it even satisfies our definition of an analysis. That is, perhaps ‘water’ expresses a complex property. And perhaps ‘liquid H₂O’ contributes to the proposition expressed by c a complex C(x) type entity that represents the property of being water and whose constituents are components of that property. But whatever its status exactly, c is not a philosophical analysis since establishing it requires scientific, and not merely philosophical, investigation. What this shows is that our characterization of analysis is not yet a characterization of philosophical analysis, since it would not in principle rule out things like c being analyses. Similarly, in my view anyway, a isn’t a philosophical analysis either, since establishing it doesn’t even require philosophical investigation.

In order to distinguish philosophical analyses from these other analyses, we must return to the claims about linguistic competence made earlier. Note that one difference between a,b and c above is that ‘brother’ in a is a category one word, ‘knowledge’ in b is a category two word, and ‘water’ in c is a category three word. This, I believe, is the key to distinguishing philosophical analyses from other analyses like a and c. Specifically, a proposition P is a philosophical analysis relative to a linguistic community c iff (i) P is an analysis; and (ii) the sentence of c expressing P has a category two word expressing the complex property being analyzed. Obviously, this makes the proposition expressed by b relative to our linguistic community a philosophical analysis, and not so for the propositions expressed by a and c.

But it is important to see the intuitive motivation behind this characterization of philosophical analyses. Recall that being competent with a category two word w requires that given any entity o that paradigmatically possesses the property P expressed by w or paradigmatically fail to possess P, were one to know all facts about o relevant to whether it possesses P or not, excluding the fact that o possesses P or the fact that o fails to possess P, one would correctly determine whether o possesses P or not. But this means that a speaker who is linguistically competent with ‘knowledge’ and who knows the relevant facts about an actual or hypothetical situation in which something is or is not a paradigmatic instance of knowledge is able to correctly determine whether the thing is or is not an instance of knowledge. By exploiting this ability and considering a wide variety of such situations, a person can, in virtue only of his linguistic competence, formulate hypotheses about the proper analysis of knowledge that explain the judgments he makes about the situations in question. Of course, he may not succeed in actually formulating a correct analysis of the relevant property. His linguistic competence guarantees only that he will make correct judgments about paradigm cases given relevant information, and not that he will hit on the correct analysis of the property that in fact explains his judgments.

The important point here is that a person can do this by considering hypothetical situations, armed only with her linguistic competence. That is, a person can do this from the armchair. Note that we now have an answer to question iii as to why philosophical analyses can in principle be produced from the armchair. They can be, because the only materials required for their production are hypothetical scenarios and linguistic competence with the category two word expressing the property being analyzed.

b. By way of contrast, suppose a linguistically competent person is considering a category one word, say ‘brother’, and the property it expresses. In virtue of her linguistic competence, she will know the components of the complex property and how they are combined to form it. That is, she will be in a position to formulate and correctly assert things like

(2) For all x, x is a brother iff x is a male sibling.

Since this is so merely in virtue of her linguistic competence, any linguistically competent person in her community will also be able to do so. Thus, things like 2 will seem trivial, unlike claims like 1.

Further, if a linguistically competent person is considering a category three word and the property it expresses, given something that paradigmatically possesses the property or fails to and the relevant facts about the situation, her linguistic competence will not allow her to correctly determine whether the thing possesses the property expressed by the word or not. Thus, she will not be able to consider a wide variety of hypothetical situations, and judge correctly whether the property in question is instantiated in those situations. Thus, absent such judgments, she will be in no position to formulate hypotheses regarding the analysis of the property in question that would explain the judgments that she can’t make! She therefore will not be able to formulate analyses of properties expressed by category three words from the armchair. In saying this, we have given an answer to question iv as to how the following differ:

a. For all x, x is a brother iff x is a male sibling.

b. For all x, x is knowledge iff x is a justified true belief.

c. For all x, x is water iff x is liquid H₂O.

A. Ackerman claims that analyses must be knowable a priori and necessarily true.

1. But as Ackerman herself points out, other examples that don’t seem to be analyses satisfy these conditions. For example:

8. For all x, x is 28 iff x is the second smallest perfect number.

2a. For all x, x is a brother iff x is a brother.

Neither of these seems to be an analysis, yet both are arguably necessary and knowable a priori. So some third condition is needed to distinguish real analyses like 1 from things like 2a and 8.

B. One property or concept P analyzes another property or concept Q iff i) P and Q are necessarily coextensive; and ii) that something is Q iff it is P is knowable a priori. Ackerman’s key idea for the needed third condition is this: iii) the claim that necessarily something is Q iff it is P can be arrived at and justified by "the philosophical example and counterexample method."

1. Ackerman [1986] describes the method as follows (my underlining):

Roughly speaking, Ackerman’s claim is that the crucial third condition required for P to analyze Q is that the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described.

C. Why Ackerman’s account isn’t correct

1. A minor worry with Ackerman’s account is cashing out the modal element in her condition: in what sense is it that the claim that necessarily something is Q iff it is P can be the outcome of the method described?

2. A much more serious problem: the problem is noted by Ackerman [1986] herself, and concerns the unexplained notion of generalization employed in her characterization of the third condition required for an analysis. Suppose that in considering the cases K counts as Q and those that she doesn’t, using a priori philosophical method I generalize from these and conclude that P analyzes Q. But now consider any R that is knowable a priori to be coextensive with Q and that is necessarily coextensive with Q. That is, consider any R that satisfies only Ackerman’s first two conditions. The question is, why doesn’t R satisfy the third condition as well and provide the analysis of Q? For all we’ve said, I could easily have generalized from K’s responses and concluded that R, instead of P, analyzes Q. After all, nothing has been said about generalization from K’s judgments about which cases are Q except that we have to capture those judgments. But since R is knowable a priori to be coextensive with Q and is necessarily coextensive with Q, supposing that R analyzes Q will explain K’s judgments just as well as supposing that P analyzes Q will. Recall that R by hypothesis is an arbitrary property that is a priori knowable to be coextensive with Q and necessarily coextensive with Q. But the whole point of adding the third condition involving generalization from K’s responses was to prevent R from analyzing Q in any case where it is a priori and necessary that R and Q are coextensive. But given any R for which it is necessary and a priori that R is coextensive with Q, we haven’t been told yet why R can’t be the outcome of generalizing from K’s responses about hypothetical cases instead of P. But then we haven’t been told why, for any R for which it is necessary and a priori that something is Q iff it is R, R might yet fail to analyze Q. But telling us this was the whole point of Ackerman’s third condition. So her account so far does not get beyond that claim that if P analyzes Q, it is necessary and a priori that P and Q are coextensive. But this is not yet an account of analyses; it is simply an obvious necessary condition on them.

3. Another serious problem: the second problem with Ackerman’s account involves a dilemma as to how precisely to understand her third condition. Either the fact that the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described is explained by or supervenes on some other relation between P and Q or it doesn’t. That is, we have this fact: the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described. Now either this fact about P and Q is explained by or supervenes on some further relation between P and Q or it doesn’t. I don’t think we can accept the latter disjunct. How could it be a brute fact, a fact without further explanation, that the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described? Surely there should be some further explanation of this fact. There should be some further relation between P and Q that explains why the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described. Okay, so suppose there is. Call this relation R. Then it seems to me that the third condition really is that P and Q stand in R, whatever R is. But then Ackerman hasn’t given us an account of analysis. She has only said this: P analyzes Q iff i) P and Q are necessarily coextensive; ii) that something is Q iff it is P is knowable a priori; and P and Q stand is some relation R that explains why the claim that necessarily something is Q iff it is P can be the outcome of the sort of philosophical inquiry described. Of course, until we are told what R is, we have no real account of analysis.

3. A final problem: Ackerman’s account has a difficulty shared by any account according to which when P analyzes Q, P and Q are distinct concepts or properties. Intuitively, when one tries to give an analysis of some concept or property Q, one seems to be trying to say what Q is and not merely trying to specify another concept or property intimately connected to Q. Just try to imagine attempting to give an analysis of reference, or moral goodness or mental representation and ask yourself whether this isn’t so. Of course, if we became convinced that no account that denied that when P analyzes Q, P and Q are distinct is workable, we could be driven to the view that we must embrace this counterintuitive claim. But I have tried to show that there is a workable account that denies this.

A. The two dimensional framework

1. Jackson and Chalmers work in a so-called two dimensional semantic framework. According to such a framework, expressions are associated with two intensions, that is, two functions from possible worlds to extensions. What they call the primary or A intension of an expression maps a possible world "considered as actual" to the extension of the term at that world. That is, consider a world w very much like ours except that XYZ fills the oceans, lakes and rivers of that world, and is what comes out of faucets and is what people drink. Now suppose that w is the actual world. That is, suppose we discover that XYZ fills our lakes, oceans etc. Then the extension of ‘water’ in w is XYZ. That is, the primary intension of ‘water’ maps the world to XYZ. But now let us consider w as a way things might have been but aren’t. Water is H₂O in the actual world, and XYZ fills the lakes and oceans of w. Then the stuff in the oceans and lakes of w isn’t water. That is, the secondary or C intension of ‘water’ maps w to H₂O, just as it maps every world to H₂O. So the primary intension of a term maps a world to the extension of that term on the supposition that the world in question is actual. The secondary intension of a term maps a world to the extension of that term on the supposition that the actual world is the actual world. Loosely put, the primary intension of ‘water’ maps a world to whatever plays the "water role" in that world and the secondary intension of ‘water’ maps every world to H₂O.

2. Now this should make clear that in the general case, knowing the secondary intension of a term requires knowing what the actual world is like in certain ways. Knowing that ‘water’ maps every world to H₂O requires knowing that in the actual world water is H₂O. Not so with a term’s primary intension. I can know that if the XYZ world is actual, then the extension of ‘water’ is XYZ without knowing what water is in the actual world.

B. Conceptual analysis

1. According to Jackson and Chalmers, conceptual analysis essentially amounts to determining what the primary intension of a term is. So a conceptual analysis of water would amount to figuring out what the extension of ‘water’ is at possible worlds considered as actual. At the actual world, it is H₂O, at the XYZ world recently discussed it is XYZ, and so on.

a. Two minor caveats.

i. Chalmers is skeptical about the possibility of finding a finite expression whose primary intension is the same as that of the expression involved in the conceptual analysis. So he is skeptical about the possibility of expressing his conceptual analyses by means of sentences like 1:

(1) For all x, x is an instance of knowledge iff x is a justified true belief.

Jackson is less skeptical in this regard.

ii. Second, Chalmers thinks there is no reason to think that the primary intensions of terms are total functions on the set of all possible worlds. For example, some worlds may be such that if they are taken to be actual, we wouldn’t know what to say about the extension of ‘water’. So for Chalmers at least, conceptual analysis may not yield any neat statements like 1, nor even primary intensions that are total functions from worlds to extensions. He thinks conceptual analysis will at any rate yield a bunch of conditionals of the following form:

8. If the actual world is thus and so, then the extension of ‘water’ is so and so (or alternatively, water is so and so).

C. Criticisms of the Chalmers/Jackson account

1. First, Jackson and Chalmers hold that the primary intensions of terms can vary from speaker to speaker. Jackson [1998] writes:

And Chalmers and Jackson [2001] write:

To say that the subjects have different conditional abilities is just to say that they associate different primary intensions with the terms. But now consider two different moral philosophers arguing about the proper analysis of moral goodness. Imagine that both are sincere, reflective and intelligent. Imagine that their analyses yield different results. That is, they entail that different things are morally right in different situations. Suppose, finally, that the philosophers have thought long and hard about the cases they disagree on and each remains firm in her opinion. If the Chalmers/ Jackson account of analysis is correct, we should at this point tell these two poor souls that they aren’t really arguing at all! It just turns out that they associate different primary intensions with the term ‘moral goodness’, just as, according to Jackson, some associate a different primary intension with ‘knowledge’ than the rest of us. But if this is right, then some of the central disputes in philosophy aren’t really disputes at all. I think that this consequence of taking the Chalmers Jackson view of conceptual analysis to be an account of philosophical analysis is very unfortunate.

2. But there is a second, more telling reason for not so taking the Chalmers Jackson view of conceptual analysis. Suppose 1 is in fact a correct analysis:

(1) For all x, x is an instance of knowledge iff x is a justified true belief.

On the Chalmers Jackson view, this would be because ‘justified true belief’ has the same primary intension as ‘knowledge’. But for terms like ‘knowledge’ and ‘justified true belief’ the primary intension is the same as the secondary intension. Whether we consider a world as actual or as counterfactual, the same things will count as knowledge. But then since ‘knowledge’ and ‘justified true belief’ have the same primary intension, they have the same secondary intension. This in turn means that the whole sentence 1 has the same primary and secondary intensions as the sentence 1a:

(1a) For all x, x is an instance of knowledge iff x is an instance of knowledge

But then the two dimensional framework of Chalmers and Jackson has no resources for dealing with the paradox of analysis. 1 and 1a share all the semantic values provided by the framework. Of course, one could introduce more resources. But then there is the danger of stipulating an ad hoc solution to the paradox of analysis. In any case, my point is that the Chalmers Jackson account of conceptual analysis by itself provides no means of dealing with the paradox of analysis. And of course, I said at the outset that for an account of philosophical analysis to be a contender, it must solve the paradox of analysis. The Chalmers Jackson view does not, and so in my view is not a contender.

References

Ackerman, Diana F., 1981, 'The Informativeness of Philosophical Analysis' , Midwest Studies in Philosophy V1

Ackerman, Diana F., 1986, 'Essential Properties and Philosophical Analysis', Midwest Studies in Philosophy XI

Chalmers, David, 1996, The Conscious Mind: In Search of a Fundamental Theory, Oxford University Press

Chalmers, D.J. and Jackson, F., 2001, ‘Conceptual Analysis and Reductive Explanation’, Philosophical Review 110: 315-361; also available on Chalmers website. The pagination of the version on the web is used in the text.

Jackson, Frank, 1998, From Metaphysics to Ethics: A Defence of Conceptual Analysis, Oxford University Press

King, Jeffrey C., 1994, 'Can Propositions be Naturalistically Acceptable?', Midwest Studies in Philosophy XIX

King, Jeffrey C., 1996a, 'Structured Propositions and Sentence Structure', forthcoming in Journal of Philosophical Logic

King, Jeffrey C., 1996b, 'Structured Propositions and Complex Predicates', forthcoming in Nous

Langford, C.H., 1942, ‘The Notion of Analysis in Moore’s Philosophy, in The Philosophy of G.E. Moore, P.A. Schilpp (ed), Open Court, La Salle, Illinois.

Moore, G.E., 1942, ‘A Reply to my Critics’, in The Philosophy of G.E. Moore, P.A. Schilpp (ed), Open Court, La Salle, Illinois.