King on Philosophical Analysis

Formulating and evaluating analyses are among the philosopher’s main activities.  A central project in ethics, for instance, is to provide an adequate analysis of social justice.  In epistemology, analyses of knowledge and justification are sought.  We are not in a position to evaluate the truth or falsity of our beliefs until we have a clear fix on their content.  The philosophical activity of formulating and evaluating analyses is aimed at clarifying the contents of our beliefs and assertions.  Careful analysis is often crucial to avoiding the fallacy of equivocation, slipping from one sense of a term to another in the same line of reasoning.  But just what is an analysis and what does a good analysis tell us?  This short essay is offered as an exposition of an account of analysis offered by Jeff King. [1] 

We need to know what sorts of things the objects of analyses are and what analyses tell us about their objects.  King takes the objects of analysis to be properties and relations.  The view of analysis to be explained here is that analyses are accounts of complex properties or relations in terms of their structure and their component properties and relations.  A crucial point here is that the objects of analyses are not terms but the properties expressed by terms.  Analysis, on this view, is not the project of associating various linguistic things.  That is, an analysis is not a mere definition of some word in terms of some other words.  We should also make clear that the objects of analysis are not conceptual where concepts are thought to be something mental.  In taking the objects of analysis to be properties and relations, we are to understand analyses as accounts of the nature of entities that are independent of our minds and language.  This view of properties and relations as entities existing independent of minds and language yet serving as the objects of thought and analysis allows for the possibility of our grasping properties and relations through our experience of their instances but failing to fully understand their specific natures.  On the proposed view, analyses are not trivial truths ascertainable just by careful reflection on one's linguistic or other beliefs.  Rather, analyses are substantive facts about the natures of the properties and relations.  We can know properties and relations in the acquaintance sense through our experience of their instances, yet lack propositional knowledge concerning their specific nature.  Proposed analyses are propositional conjectures about the nature of properties and relations. 

We will make use of the main elements of King's account, but we need not commit to the full framework, which includes an independently motivated account of propositions.  We start with the assumption that some properties and relations are complex and have other properties and relations as constituents.  Proposed analyses of dispositions to be discussed here will be given in what King calls "orthodox form":

(2)  ("x)(Px iff Cx)

(2) gives the orthodox form of an analysis of a property.  An analysis of a relation would have a predicate with multiple free variables expressing the relation to be analyzed to the left of "iff" and a complex expression having just those variables free to the right of "iff."  Other formulations are often used informally in giving analyses.  We might say, for instance, "A disposition is a property such that. . ." or "To be a disposition is to be a property such that. . ."  But in explicating the notion of analysis in terms of the form of (2), we avoid having to address the semantics of "is" or "to be."  In giving analyses in the form of a bi-conditional, however, we should bear in mind that not every true bi-conditional expresses an analysis.  The further condition, again, is that the expression to the right of "iff" represents properties and relations that are constitutive of the analysandum.  We will also take the structure of the complex analysandum to be partially represented by the syntactic relations holding among component property and relation expressions in the analysans.

In an analysis of a property given in orthodox form, the predicate to the left of "iff" will express the analysandum, the property that is the object of analysis.  The predicate to the right of "iff" will be a complex expression having only x free which includes predicates expressing properties and relations that are components of the analysandum.  The analyzing predicate to the right of "iff" does not express the same property as the predicate to the left.  But it can be said to "represent" or "stand for" the analysandum by having elements that express properties and relations constitutive of it and by representing those component properties and relations as standing in a complex relation that reveals some of the structure of the property or relation that is the object of analysis.  That structure revealing relation is represented by the syntax of the analyzing predicate.  I will have more to say about how the syntax and semantics of analyzing expressions represent the structure of their analysandum shortly.

A primary motivation for King's treatment of analysis is to provide a solution to the problem of analysis.  That the predicates to the right and left of "iff" in an analysis have distinct semantic values is essential for King's resolution of the problem of analysis.  It is commonly held that in a correct analysis, the expression to the left of "iff", indicating the object of analysis or the analysandum, and the expression to the right of the "iff", expressing the analysans, "express the same thing", or "mean the same thing."  But if this is so, then the analysis will be synonymous with the biconditional that results from substituting the predicate to the left of "iff" for the analyzing predicate to the right of "iff" in the analysis.  For instance, if the expressions to the right and left of "iff" in an analysis express the same thing, then (3) and (4) below should be synonymous:

(3)  x is a bachelor iff x is an unmarried adult male.

(4)  x is a bachelor iff x is a bachelor.

But (3) is an analysis and (4) is not.  (3) but not (4) tells us something about what it is to be a bachelor.  If the expressions to the right and left of "iff" in (3) express or mean precisely the same thing, then we are apparently unable to explain how analyses can be informative.  But analyses are informative.  An analysis of gold as the element having atomic number 79 can hardly be deemed uninformative.  Nor is this result acceptable in the case of typical philosophical analyses.  This is the problem of analysis.  A solution to the problem of analysis requires some account of the semantics of analyses that will provide a distinction between the meanings, that is, the semantic values, of the predicates to the right and left of "iff" in an analysis.  King offers an independently motivated account of propositions as complex structured entities that does just this.  I will not present the full account here.  But for the sake of elucidating King's treatment of the problem of analysis, I will offer a brief summary. 

Propositions are complex structured entities on King's account.  When a sentence expresses a proposition, the proposition is structured by what King calls the "propositional relation."  The propositional relation is the composition of the syntactic relation that holds among the lexical items of the sentence and the semantic relations that connect lexical items with their semantic values.  A sentence's underlying syntactic structure is not always reflected in its surface structure, but we will not pursue these details.  The syntactic inputs to semantic analysis are the lexical items of a sentence structured by its syntactic relation.  The proposition expressed by a sentence consists of the semantic values of its lexical items standing in a propositional relation that is the composition of the syntactic relation holding among the sentence's lexical items and the semantic relations that hold between the lexical items and their semantic values.  The semantic values for names, predicates and logical terms are, respectively, bearers, properties or relations, and logical relations.  So, we can represent the structure of a proposition with a syntactic tree with its terminal nodes extended to represent the semantic relations holding among lexical items and their semantic values.

In an analysis, the predicate to the left of "iff" has as its semantic value, the property or relation that is the object of analysis.  The complex predicate to the right of "iff" can be said to "represent" the analysandum, but it does not contribute that property or relation as a semantic value to the proposition expressed by the analysis.  Rather, the semantic value of the predicate to the right of "iff" in an analysis is a "complex entity" that consists of various properties and relations that are components of the analysandum standing in a complex propositional relation.  In this manner, King's account marks a difference in the semantic values of the predicates to the right and left of "iff" in an analysis.  While their semantic values differ, they are related in a way that allows us to take the analysans as elucidating the nature of the analysandum.

So, for example, the property of being a bachelor will be analyzed as follows: 

(5) x is a bachelor iff x is unmarried, x is adult and x is male.

The predicate on the right of "iff" in (5) is composed of predicates that express properties and relations that are components of the property of being a bachelor.  The component properties are the properties of being unmarried, being adult and being male.  A component relation is also represented.  The three component properties stand in a three-place conjunctive relation.  This constituent relation gives part of the structure of the property of being a bachelor.  The three-place conjunctive relation gives the structure of the property in so far as it consists of the constituent properties of being unmarried, adult and male.  But the conjunctive relation expressed in this analysis does not reveal all of the structure of the complex property of being a bachelor.

Not all of the properties and relations constitutive of the analysandum need to be explicitly mentioned in the analyzing predicate.  Nor must all of the structure of the analysandum be explicitly represented in the analyzing predicate.  The three-place conjunctive relation does not give the full structure of the property of being a bachelor, but only that part of the structure that holds between the constituent properties of being adult, being male, and being unmarried.  The properties and relations mentioned in the analyzing predicate may be complex themselves.  Properties and relations constitutive of these are also constitutive of the object of analysis.  In the case of the analysis of bachelorhood the components of the complex property of being unmarried, for instance, are also components of the property of being a bachelor.  A more refined analysis of the property of being a bachelor could be given in terms of the property of being adult, being male, and the properties and relations that are components of the property of being unmarried.  Alternatively, the property of being a bachelor could be analyzed in terms of just the properties of being unmarried and being a man, where the property of being a man is just the complex property composed of the property of being adult and the property of being male conjunctively related.  So, a complex property or relation may have a large number of correct analyses varying in the degree to which they make explicit the constituents and structure of the object of analysis. In any case, all of the analysandum's constituent properties and relations must either be, or be components of, the properties and relations expressed or syntactically represented in the analyzing predicate.  We may say that analyses are complete accounts of a property or relation's nature in just this sense.  But an analysis of a complex property or relation need not be completely or exhaustively detailed in the sense of having lexical items that express, or syntactic structure that represents every property and relation constitutive of the analysandum. 

King distinguishes three classes of words according to what is required for linguistic competence.  For the first class, linguistic competence requires that one be able to articulate the properties and relations that are the components of the complex object of analysis.  "Bachelor" is an example of a word belonging to this class.  For the second class, linguistic competence requires that one's usage be guided by the presence or absence of the constituent properties.  Philosophical analyses are typically aimed at explaining the nature of the complex properties expressed by predicates belonging to this class.  A competent speaker's usage of the term "knowledge", for instance, is guided by the presence or absence of the constituent properties and relations, but a competent speaker need not be able to specify just what those constituent properties and relations are.  That competence in using such words does not require the ability to articulate constituent properties is indicated by the competent speaker's intuitive inclination to reject Gettier cases as instances of knowledge.  A competent speaker need not be able to articulate what condition C must be met beyond an agent's belief being true and justified in order to correctly judge that an agent's belief that P in a Gettier case does not count as an instance of knowledge. 

For the third class, linguistic competence requires neither knowledge of component properties or relations nor that usage be guided by the presence or absence of these.  King suggests that scientific analyses of natural kinds concern words belonging to this class.  A person need not know the atomic number of Gold nor understand how that figures in an analysis of the nature of gold in order to be competent with the word "gold".  Neither must a speaker who is competent with the word "gold" be able to distinguish gold from fool's gold.

An important feature of this account of analysis is that analyses are not trivial or merely conceptual truths.  Discovering the correct analysis of a property expressed by words of the second or third class involves substantive investigation and the resulting analyses will, as they should, be informative.