Philosophy 2 Week 2 Essay Quiz

Analysis, Language, and Concepts: The Second Paradox of Analysis

Author(s): Felicia Ackerman

Source: Philosophical Perspectives , 1990, Vol. 4, Action Theory and Philosophy of Mind (1990), pp. 535-543

Published by: Ridgeview Publishing Company

Stable URL: http://www.jstor.com/stable/2214203

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

and Wiley are collaborating with JSTOR to digitize, preserve and extend access to Philosophical Perspectives

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

Philosophical Perspectives, 4

Action Theory and Philosophy of Mind, 1990

ANALYSIS, LANGUAGE, AND CONCEPTS: THE SECOND PARADOX OF ANALYSIS'

Felicia Ackerman Brown University

Center for Advanced Study in the Behavioral Sciences

In an earlier paper (Ackerman, forthcoming; see also Ackerman 1981 and 1986), I distinguished two paradoxes of analysis and offered a solution to one. This paper offers a solution to the other and also touches on some general issues about analysis, language, and concepts.

The two paradoxes of analysis can be illustrated and distinguished by considering the following propositions.

(1) To be an instance of knowledge is to be an instance of

justified true belief not essentially grounded in any falsehood.

(2) To be an instance of knowledge is to be an instance of knowledge.

(3) A correct analysis is given by saying that to be a brother is to be a male sibling.

(4) A correct analysis is given by saying that to be a brother is to be a brother.

The first paradox arises from the relation between (1) and (2). (1), if true, illustrates an important type of philosophical analysis. For purposes of illustration, I will assume (1) is a correct analysis. To suppose that (1) gives a correct analysis, however, leads to the following notorious problem. If the concept of justified true belief not essentially grounded in any falsehood is the analysans of the

concept of knowledge, it would seem that they are the same concept and hence that (2) would have to be the same proposition as (1). But

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

536 / Felicia Ackerman

propositions (1) and (2) seem clearly diverse; for example, (2) is trivial

(in the sense that understanding it entails believing it), while (1) can

be informative and in fact was only recently arrived at by philosophers.

This is the paradox philosophers generally call "the" paradox of

analysis. But there is another paradox of analysis that can be gleaned

by reading classic writings on analysis, such as Moore, 1952 and Langford, 1952. This second paradox of analysis arises from the relation between (3) and (4). If (3) is true, it would seem that the

concept of being a brother would have to be the same concept as

the concept of being a male sibling and that (4) would also have to

be true and in fact would have to be the same proposition as (3).

Yet (3) is true and (4) is false.

Both these paradoxes rest upon the assumptions that analysis is

a relation wholly between concepts rather than one between entities

of other. sorts, such as linguistic expressions, and that in a correct

analysis, analysans and analysandum will be the same concept. Both these assumptions are explicit in Moore, 1952. But some of Moore's remarks hint at a solution to the second paradox-that a statement

of an analysis is a statement partly about the concept involved and partly about the verbal expressions used to express it. (Moore, 1952, p. 283.) He says he thinks a solution of this sort is bound to be right,

but fails to suggest one because he cannot see just how the analysis can be even partly about the expressions.

In this paper I will suggest a way and thus offer a solution to the

second paradox. It is important to see that this sort of approach cannot possibly provide a solution to the first, however. The two

paradoxes must be clearly distinguished. This can be done by means of the Frege-inspired sense-individuation condition, which is the condition that two expressions have the same sense if and only if they can be interchanged salva veritate whenever used in propositional attitude contexts. If the expressions for the analysans

and the analysandum in (1) met this condition, (1) and (2) would not raise the first paradox, but the second paradox arises regardless of

whether the expressions for the analysans and the analysandum meet

this condition. The second paradox is a matter of the failure of such expressions to be interchangeable salva veritate in sentences involving such contexts as 'A correct analysis is given by'. Thus, a solution aimed only at such contexts can solve the second paradox. This is clearly false for the first paradox, however, which will apply

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

The Second Paradox of Analysis / 537

to all pairs of propositions expressed by sentences in which

expressions for pairs of analysanda and analysantia raising the first paradox are interchanged. It seems obviously unacceptable to hold that the proposition that John knows that the moon is not made of green cheese is partly about language because it can be believed

without believing the proposition that John has justified true belief

not essentially grounded in any falsehood that the moon is not made of green cheese.

My solution to the first paradox (Ackerman, 1981, 1986, and

forthcoming) rested on denying Moore's view that analysans and analysandum have to be the same concept in a correct analysis.

Instead, I argued that in analyses raising the first paradox, the analysans and analysandum are diverse concepts that are necessarily and a priori knowable to be coextensive and are further related in

a special epistemic way I spelled out. Such a solution seems attractive

in cases such as (1) and (2), for the concepts of knowledge and justified true belief not essentially grounded in any falsehood seem on the face of it to be quite different. The first is readily grasped, the second fairly difficult; the first is employed in the thoughts of many people,

the second in those of relatively few, perhaps only professional philosophers and their students. But this sort of solution seems precluded where we have pairs of synonymous expressions that are

interchangeable salva veritate whenever used in propositional attitude contexts, and thus where analysis raises only the second paradox and.not the first at all. Purely for the sake of illustration, I will assume throughout this paper that 'brother' and 'male sibling' are such a pair of expressions, and that the concept of being a brother is identical with the concept of being a male sibling. (Nothing in my argument hinges on this particular example, and readers who dislike

the example can substitute one of their own.) This makes it reasonable

to suppose, with Moore (1952), that (3) is partly about words, while rejecting the view that it is wholly about the verbal expressions 'brother' and 'male sibling' and not about the concept at all-a view Moore opposes by arguing as follows.

For what are you doing [in saying simply that the

expressions 'brother' and 'male sibling' have the same

meaning]? You are merely asserting, with regard to two verbal expressions, that...the meaning they have is the same. You are not mentioning the meaning of either, or saying

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

538 / Felicia Ackerman

what the meaning of either is; but are merely making a

statement, which could be completely understood by a

person who had not the least idea what either expression meant. A man might point out to me two expressions in a

language of which I was completely ignorant and tell me

that they had the same meaning, without telling me what

they meant. So far as he was merely telling me that they

had the same meaning, I should completely understand what he told me-namely that those two expressions had the

same meaning.... But if this were all he was doing, he would

not have told me anything at all about any concept or idea,

which either of the expressions expressed and would

therefore certainly not have been giving me an analysis of any concept....(Moore, 1952, pp. 280-1, italics in original.)

But he goes on to say

An obvious suggestion to make is that, if you [assert (3)], you are making a statement both about the concept brother and also about the two verbal expressions used, which

would explain why this statement is not the same statement

as [(4)]. But this suggestion would be compatible with its

being the case that [(3)] is merely a conjunction of the assertion "The verbal expression 'x is a brother' has the same meaning as the expression 'x is a male sibling"' with some other assertion which is merely an assertion about the

concept "x is a brother" and not an assertion about any verbal expression. But I do not think this can possibly be the case: what would the second assertion in this supposed

conjunction be? (Moore, 1952, p. 283, italics in original.)

An obvious reply to Moore's final question here would be to say

the second conjunct is simply that the concept of being a brother

is identical with the concept of being a male sibling. Then (3) would

be explicated as

(5) A correct analysis is given by saying that the verbal

expression 'x is a brother' has the same meaning as the verbal expression 'x is a male sibling' and the concept of

being a brother is identical with the concept of being a male sibling.

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

The Second Paradox of Analysis / 539

But then the original problem rearises, as the identity of the concept

of being a brother and the concept of being a male sibling means

(5) is the same proposition as

(6) A correct analysis is given by saying that the verbal

expression 'x is a brother' has the same meaning as the

verbal expression 'x is a male sibling' and the concept of

being a brother is identical with the concept of being a

brother.

But (6), like (4), seems clearly false.

The inadequacy of (5) should not make us reject the general view that (3) is partly about words and partly about concepts. Moore is

obviously right in his above quoted remark that the general position

that (3) is partly about words and partly about concepts is compatible with its being the case that (3) is "merely a conjunction of the assertion

'The verbal expression 'x is a brother' has the same meaning as the

expression 'x is a male sibling" with some other assertion which is

merely about the concept "x is a brother" and not an assertion about any verbal expression." (Moore, 1952, p. 283, italics in original.) But it obviously does not follow that every particular proposal that makes (3) partly about words and partly about concepts is compatible with

saying (3) is merely a conjunction of this sort and hence open to the

sort of objection facing (5). Here is such a proposal that is not.

(7) A correct analysis is given by saying that the verbal expression 'x is a brother' expresses the same concept as is expressed by the conjunction of the verbal expressions 'x is male' when used to express the concept of being male and

'x is a sibling' when used to express the concept of being a

sibling.

(7) pairs each constituent concept in the analysans separately with the verbal expression that expresses it, so (7) cannot be paraphrased

as a conjunction that leads to the problems of (5). I suggest (7) as the solution to Moore's question of how analysis can be partly about words and partly about concepts and thus as an account of (3) that does not entail anything raising the problems of (4) and (5) and so solves the second paradox of analysis. The rest of this paper will discuss implications of this solution and possible objections to it.

An important point about (7) is as follows. Stripped of its philosophical jargon ('analysis', 'concept', 'x is a ...'), (7) seems to state

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

540 / Felicia Ackerman

the sort of information generally stated in a definition of the verbal

expression 'brother' in terms of the verbal expressions 'male' and

'sibling', where this definition is designed to draw upon the listeners' antecedent understanding of the verbal expressions 'male' and

'sibling' and thus to tell listeners what the verbal expression 'brother'

really means, instead of merely providing the information that two

verbal expressions are synonymous without specifying the meaning

of either one. Thus, the solution to the second paradox seems to make the sort of analysis that gives rise to this paradox a matter of

specifying the meaning of a verbal expression in terms of separate

verbal expressions already understood and saying how the meanings

of these separate, already-understood verbal expressions are

combined. (This corresponds to Moore's intuitive requirement that an analysis should both specify the constituent concepts of the

analysandum and tell how they are combined. See Moore, 1952, p. 284.) But is this all there is to philosophical analysis?

To answer this question, we must note that, in addition to there

being two paradoxes of analysis, there are two types of analysis that are relevant here. ([here are also other types of analysis, such as reformatory analysis, where the analysans is intended to improve

on and replace the analysandum. But since reformatory analysis involves no commitment to conceptual identity between analysans

and analysandum, reformatory analysis does not generate a paradox of analysis and so will not concern me in this paper.) The important distinction between the two types of analysis I am concerned with

is the one mentioned earlier in distinguishing the two paradoxes

themselves: the expressions for some pairs of analysanda and analysantia are interchangeable salva veritate whenever used in

propositional attitude contexts, while the expressions for other pairs

are not. This corresponds to the rough intuitive distinction between

analyses that cannot be informative to people who understand all

the words used in expressing the analysis and analyses that can be.

Both paradoxes apply to the pairs that are not interchangeable, but

only the second applies to those that are. I have already mentioned that elsewhere (Ackerman, 1981, 1986, and forthcoming), I have

offered a solution to the first paradox, based on the view that in the cases where this paradox applies, analysans and analysandum are diverse, but closely related, concepts. This will also work as a solution to the second paradox for cases where the first paradox applies,

making the sort of analysis to which the first paradox applies purely

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

The Second Paradox of Analysis / 541

a relation between concepts. So (7) is intended as an account of (3)

only on the assumption that the concept of being a brother is identical

with the concept of being a male sibling and the verbal expressions 'brother' and 'male sibling' are synonymous and are interchangeable

salva veritate whenever used in propositional attitude contexts.

Yet some objections remain. For one thing, the solution is not

completely satisfying intuitively. (How many solutions ever are?)

Specifically, there is still a temptation to suppose we can perform a purely conceptual process of breaking down a concept into its parts and saying how they are combined (as in (3)) and that this will be

different from merely repeating the concept (as in (4)). But this is

precisely what gives rise to the second paradox. Some additional

uneasiness here may lie in doubts about whether the verbal expressions 'brother' and 'male sibling' actually are interchangeable

salva veritate whenever used in propositional attitude contexts-

something I indicated that I was assuming for the sake of illustration. There are writers nowadays who deny that synonymy requires such

interchangeability (e.g., Burge, 1978). But in this paper I am concerned only with cases, whatever they are, that raise the second paradox without raising the first paradox, for example, when one

term is always thought of as an abbreviation or shorthand for the

other and is always used and understood this way. (Of course, my whole sort of approach is deeply opposed to Burge's, but I cannot go into all our differences here.) In such cases it does seem that the information stated by such sentences as the one in (3) (over and above information about the concept of an analysis) is linguistic information.

These considerations also explain a related apparent drawback of

my solution, which can be illustrated by considering the relation between

(8) A correct analysis is given by saying that to be a

fortnight is to be a period of two weeks

and

(9) Une analyse correcte est donnee par dire qu'etre une

quinzaine c'est etre une periode de deux semaines.

Let us assume for the sake of argument that 'fortnight' and 'period

of two weeks' are synonymous and are interchangeable salva veritate whenever used in propositional attitude contexts, as are 'quinzaine' and 'periode de deux semaines'. (I have changed the example from

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

542 / Felicia Ackerman

that of (3) because French lacks a one-word synonym for 'sibling'.)

Intuitively, one might want to suppose that (8) and (9) are the same

proposition. On my account, however, they are not, since (8)

mentions English words and (9) mentions French ones. This is an

inevitable result of the considerations already mentioned, but my

account does allow for (8) and (9) to be closely related in an obvious

way.

Another possible objection is as follows. Although the sentence I use in (7) does not use either the expression 'male sibling' (i.e., the two words conjoined) or the expression 'brother,' an objection may

seem to arise from considering the relation among (7),

(10) When the verbal expression 'x is a brother' is used to

express the concept of being a brother, a correct analysis is given by saying that the verbal expression 'x is a brother'

expresses the same concept as is expressed by the

conjunction of the verbal expressions 'x is male' when used to express the concept of being male and 'x is a sibling'

when used to express the concept of being a sibling,

and

(11) When the verbal expression 'x is a brother' is used to

express the concept of being a male sibling, a correct

analysis is given by saying that the verbal expression 'x is a brother' expresses the same concept as is expressed by the conjunction of the verbal expressions 'x is male' when used to express the concept of being male and 'x is a sibling'

when used to express the concept of being a sibling.

On our assumption that the concept of being a brother is identical with the concept of being a male sibling, (10) and (11) are the same proposition. But (7) entails (10), hence (11).

This is not really a problem, however. (11) does not have the unacceptable qualities of (4) and (6). (4) and (6) are clearly unacceptable because neither comes close to the intuitive desideratum of specifying the "parts" of a concept and saying how they are combined. But (11) does. I think the apparent oddity of (11) is a pragmatic one. It would be odd to use the sentence I have used in (11) when talking to a single audience, because if the audience understood the 'when' clause (and had elementary logical acumen), the remainder of the sentence would be superfluous, except to give

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms

The Second Paradox of Analysis / 543

information about the concept of an analysis. But this does not mean (11) would be false.

Notes

1. This paper was completed while I was a Fellow at the Center for Advanced Study in the Behavioral Sciences/National Endowment for the Humanities Fellow (Grant #FC10060-85). The support of both institutions is gratefully acknowledged. I also thank Ernest Sosa and James Van Cleve for helpful discussions.

References

Ackerman, D. F.: 1981, 'The Informativeness of Philosophical Analysis,' in P. French, et al. (eds.), Midwest Studies in Philosophy, vol. 6, University of Minnesota Press, Minneapolis, pp. 313-320.

Ackerman, D. F.: 1986, 'Essential Properties and Philosophical Analysis,' in P. French, et al. (eds.) Midwest Studies in Philosophy, vol. 11, University of Minnesota Press, Minneapolis, pp. 304-313.

Ackerman, D. F.: Forthcoming, 'Analysis and Its Paradoxes,' in Proceedings of the 1985 Israel Colloquium for the History, Philosophy, and Sociology of Science.

Burge, T.: 1978, 'Belief and Synonymy,' Journal of Philosophy, 75, 119-138. Langford, C. H.: 1952, 'The Notion of Analysis in Moore's Philosophy,' in

P. Schilpp (ed.), The Philosophy of G. E. Moore, Open Court, La Salle, IL, pp. 321-342.

Moore, G. E.: 1952, 'Analysis: A Reply to My Critics,' in W. Alston and G. Nakhnikian (eds.), Readings in Twentieth Century Philosophy, Free Press of Glencoe, London, 1963, pp. 279-285.

This content downloaded from ������������128.195.203.196 on Mon, 03 Aug 2020 18:07:22 UTC������������

All use subject to https://about.jstor.org/terms