By Rick Grush
Strawson's Individuals, Chapter 6: Subject and Predicate (2): Logical Subjects and Particular Objects.
I. The introduction of particulars into propositions.
1. The introduction of a particular into a proposition requires
knowledge of an empirical fact: the introducing of a universal
does not.
Term introduction is a general description applying both to subject- and predicate-expressions. But both ways of introducing require identification. The expression must indicate which term is introduced.
S first examines the conditions for introducing a particular (or referring). There are three conditions for a successful reference-attempt: i) there is a particular the speaker is referring to, ii) there is a particular the hearer takes the speaker to be referring to, and iii) those particulars are the same.
Regarding (i), this means that if the speaker uses a description, there should be something answering to it. If it is a name the speaker uses, then the speaker should be prepared to substitute a description. But these are not enough, for the P (particular) must be unique. That is, the speaker, in order to make reference to a particular, must not only know a description, but know that it applies uniquely. This is what it means for the speaker to have *a* P in mind. If in fact, any description the speaker could give fails to uniquely apply, then the speaker failed to make a genuine reference. Since the question whether there is a unique P answering to some description is an empirical fact, then both for the speaker and the hearer (as the conditions for (ii) are similar to those for (i)) knowledge is required, in some sense, of an empirical fact.
Finally, note that condition (iii) does not require that the uniquely-applicable descriptions the speaker and hearer use are the same description, only that they uniquely apply to the same P, regardless of what they are.
But there are no parallel empirical conditions on the introduction of a universal.
S considers several objections to this last claim.
First, in order to successfuly introduce a universal with an expression 'phi', it must be the case that both the speaker and hearer understand the import of propositions made with it, and this is an empirical fact. S's reply is that this is not parallel, because the required empirical fact in this case is a fact about *language*.
Second, sentences using 'phi' would not be significant unless a large portion of such sentences actually, and accurately, described states of affairs in which something was phi. Two replies to this: i) these are two very different sorts of empircal fact: one quite definite, and the other quite indifinite. ii) The facts are of a different sort in the following sense: the empirical fact required for P identification must not only obtain, but be KNOWN to obtain, in order for identifying reference to be made. The empirical fact(s) required for U introduction are historical facts, if they are facts at all, which explain how an expression got its meaning. But once that meaning is established, the speakers need not have any beliefs, true or otherwise, about the facts which established that meaning.
p. 185-6:
"The identifying intoduction of either a particular or a universal into discourse entails knowing what particular or universal is meant, or intended to be introduced, by the introducing expression. Knowing what particular is mean entails knowing, or sometimes -- in the case of the hearer -- learning, from the introducing expression used, some empirical fact which suffices to identify that particular, other than the fact that it is the particular currently being introduced. But knowing what universal is meant does not in the same way entail knowing any empirical fact: it merely entails knowing the language."
One might rephrase this as: mere competence with the language is all that is required to introduce Us successfully, but mere competence with the language does not suffice to successfully introduce Ps. Compare to this Evans' remark in VR ch. 11, p. 373 n. 1:
"There are difficulties in assimilating proper name to symbols belonging to a language (like 'red' or 'limp'), just because the groups of peope who know these expressions are often so small. (There is clearly *some* useful notion of the English language according to which proper names are not parts of it, but at best parts of particular ideolects.)"
He closes the section with a qualification, which is that noncannonical introduction of universals can require empirical knowledge. E.g. 'wise' vs. 'the characteristic most commonly attributed to Socrates'.
2. Affinity between the grammatical and categorial criteria for subject- and predicate-expressions in part explained by a mediating distinction between 'completeness' and 'incompleteness'. Once the fundamental association between subject-predicate and particular-universal distinctions is established, further extensions of the former may be explained by analogy, etc.
Call, then, the difference between expressions introducing Ps and Us a difference between expressions which require knowledge of an empirical fact to be successful, and those which do not (EEs (empirical -expressions) and NEs (nonempirical-expressions) for short.)
EEs, S claims, have a sort of completeness or sufficiency which NEs lack. EEs carry 'a weight of fact' in introducing their terms, while NEs carry no such wieght, and can only *help* to carry a fact when coupled to an EE.
S claims that the grammatical criterion can be seen as a manifestation of this deeper fact. ( The grammatical criterion, recall, was that predicate-expressions carried the assertive-propositional marking, esp. the finite verb form, which marks it as explicitly incomplete -- which marks it as something which i) is clearly meant to be part of something which expresses a fact, and yet ii) is incomplete exactly because it does no such thing on its own. The predicate expression carries the symbolism which demands completion into an assertion.)
Thus the new criterion of difference is that between those expressions which 'present facts in their own right' (and are in that sense complete), and those which do not, and hence are incomplete wrt presenting a fact.
And as the last section showed, the new criterion, call it the completeness criterion, harmonizes with the categorial criterion. Again, simply, because particular-expressions are complete, and universal expressions are not, on the currrent account of completeness.
Because the completeness criterion underlies both the grammatical and the categorial criteria, it explains the consistent association of the subject-predicate distinction with the particular-universal distinction.
S closes by discussing two problem cases.
First, universals can be used as subjects, e.g. 'generosity is a virtue'. Thus we have an NE acting as a subject, which is supposed to be the part of the expression which carries the burden of fact. S's reply here, which I confess I do not see as adequate, is that such statements work because of an analogy between the way particulars fall under universals and the way subordinate universals fall under superordinate universals.
Second, predicates such as 'being married to John' have parts which are EEs, and hence would seem to carry some empirical weight in the required sense. S's reply here is that such expressions, though they contain parts which would carry empirical weight, carry no weight by themselves as a whole. 'John' carries an empirical presupposition in order to work correctly. All 'being married to John' presupposed is that someone is or is not married to John, and this is no empirical presupposition.
BUT, does this expression introduce anything understandable if the speaker or hearer do not know who John is? It still seems to me that there is some genuinely empirical presupposition for the proper operation of this expression. Maybe this is a point of discussion.
3. Further explanations of the idea of 'completeness': the presuppositions of expressions introducing particular terms.
S says that the notion of 'completeness' is still not characterized
very well. That is the task of this section.
There may be different things meant by 'completeness' in different sorts of case. For instance with a demonstrative, such as 'That man there can help you', the empirical requirement for referring to that man is that there be exactly one man there at whom I am pointing.
But the case for proper names is not so clear. For the empirical fact cannot be that there is one person who has that name, because that is typically false. And even if there is some fact which uniquely fixes the referent of the expression, it is not necessarily the case that this fact is really presupposed by the proper functioning of the expression.
S goes on to give some account of the empirical presuppositions of the successful use of a proper name. These are that there should be exactly one person of whom 'a reasonable portion' of the facts commonly associated with the name are true. This is an area which Evans, following Kripke, seems to have greatly advanced the theory of proper name reference over the view expressed here by Strawson (and similar views current at the time).
He then basically punts on the question of spelling out the presuppositions underwriting various sorts of referring expressions, claiming that it is not his job to do that. All that is rquired is that there be such presuppositions for each such expression.
He closes by considering one final objection to his theory: The account is circular, because you (S) claim that in order to introduce a particular into a proposition, certain conditions must hold, namely identifying facts. But these facts themselves presuppose the ability to identify particulars, either directly, or, indirectly, because they at best presuppose an ability to quantify, which can only happen when one has a capacity to identify particulars.
The reply is that his theoy is concerned only with the conditions of introducing a particular particular (so to speak) into a proposition on some particular occassion, and not an account of the general capacity of particular introduction. As such, his account is not circular, because the presupposed facts for introducing a particular X, while they may depend on identifying particulars, won't require a prior identification of X.
Part 2 of this chapter is aimed at the more general question.
4. Consideration and rejection of a simplified form of the above
theory.
S considers an alternate proposal which appears to provide the same reslts as his own more complicated theory. The thoery is thatparticular-expressions can be replaced by bound variables of existential quantification. (This is what Russell famously did with definite descriptions, e.g.) S claims that on his own account, the final presuppositions, when one has ferretted out all the presuppositions behind the presuppositions of a P-expressions, will have such quantified axpressions. The difference is that the present proposal sees such Q-expressions as an immidiate analysis of P-expressions.
S ahows how this proposal accounts for the completeness of particular-expressions (subjects) and the incompleteness of universal-expressions (predicates). The idea is that a definite description is in effect a mini-proposition (There is exactly one thing which is F.') which is simply conjoined with a relative pronoun-expression (...and which...) to make a subject-expression. There is no similar decomposition of the predicate-expression possible.
S's response to this is that this program is subject to exactly the sort of circularity which he defended his proposal against in the previous section. That is, it claims to be a general program for the elimination of subject-expressions, but the apparatus it employs (quantification) presupposes the possibillity of non-quantifications subject-expressions.
II. The Introduction of particulars into discourse.
5. Can the above account of the conditions of introducing particulars into propositions be supplemented with an account of the conditions of introducing particulars into discourse? The conditions of success of any such attempt.
S summarizes the position so far, p. 198:
"One introduces a particular into a proposition if one makes an identifying reference to that particular in that proposition. In the first part of this chapter I have discussed the conditions of the introduction of particulars into propositions. The outcome of this discussion was, summarily, the doctrine that every introduction of a particular carried a presupposition of empirical fact. The propositions of fact thus presupposed could be thought of, without circularity but not without regression, as themselves involving the introduction of (identifying reference to) particulars, as well as quantification over particulars; and the ultimately presupposed propositions of fact could be thought of, without either circularity or regression, as involving quantification over particulars, though not the introduction of (identifying reference to) particulars."
He now wants to discuss the conditions of particular introduction in general, as opposed to the conditions of introduction for some specific particular. Call the latter p-introduction1 (or PI1), and the former, p-introduction2 (or PI2).
S feels that a theory of PI2 would be good, not only because his defense against the charge of circularity for PI1 in the first part of the chapter required him to claim explicitly that it was not a theory of PI2, but for general reasons of completeness as well.
In order for a theory of PI2 to avoid circularity, it must not presuppose facts which either quantify over particulars or introduce1 them.
Distinguish two types of presupposition, parallel to the two sorts of introduction, presupposition1 and presupposition2 (P1 and P2). PI1 involves the P1 of some propositions. PI2 involves the P2 of facts, which S characterizes as a 'condition of there being any propositions at all into which particulars of that kind are introduced1.'
He then considers an objection: How can a theory of PI2 aid a theory of PI1? A theory of PI2 cannot I1, or quantify over, any particulars of the class whose introduction it aims to explain. This equally prohibits I1ing any sortal universals of the type appropriate to particulars of that kind. Therefore P2 facts provide the conditions not only for PI2, but for SUI2 (where the SUs are sortal universals of the type appropriate to the particulars under discussion). So, a theory of PI2 lacks the particular/universal asymmetry characteristic of a theory of PI1.
>I don't myself see why this is an objection to S's program. Perhaps this will be a useful topic of discussion.
S replies that just because a theory of PI2 cannot P2 SUs of the appropriate sort, it does not follow that a theory of PI2 will be a theory of SUI2.
Furthermore, here is how an account of PI2 can supplement an account of PI1: Consider a case where a dependent particular is introduced as attributively tied to an independent particular, e.g. 'The death of Socrates'. In such a case, introduction of dependent particulars presupposes2 facts of a certain kind, e.g. 'Socrates died'. Such facts as these, which employ neither particulars of the kind under discussion (dependent particulars), nor SUs of the type appropriate to them, do employ characterizing universals (e.g. *dying*). Such a fact is P1ed by the PI1 of such a dependent particular, but is also P2ed by the PI2 of that class of particulars.
But in general, we shold not expect that the P1ed propositions will also be the P2ed propositions.
p. 201:
"The general nature of the connexion between the theories is, rather, this: if we accept the theory of presuppositions1, then saying that the existence of a certain class of particular-introducing expressions presupposes2 the existence of a certain class of facts involves claiming that we can think of some introduction1 of a particular of the relevant class as presupposing1 a fact of the presupposed2 class."
The previous example of *dying* and deaths was of a sort where this was particularly clear. But in general, a good theory of P2s will make a theory of P1s much easier. He gives an example, p 201:
"...propositions about nations cannot be reduced to propositions about men; but propositions about men are presupposed2 by propositions about nations."
The problems will come when we get to the presuppositions2 of introducing that class of particulars on whose introduction2, introduction2 of all other particular classes depends. p. 202:
"At the last stage, it means that the universals contained in the presupposed2 propositions must not function as sortal or characterizing universals at all. This seems indeed a severe requirement. Where are we to find propositions which both have this character and are adequate to supply a basis for the introduction of the fundamental kinds of particular?"
6. Feature-concepts and sortal universals: the introduction of basic particulars involves the adoption of criteria for reidentification.
Universal which fit the bill (neither sortal nor characterizing) are feature-universals, or FUs, as used in feature-placing statements (FPSs). (Didn't think I was gonna stop with the acronyms, did you?)
Examples:
Now it is raining.
Snow is falling.
There is goal here.
None of these FUs provides, or presupposes, a way of individuating particulars. This is true even though such expressions can figure in such ways as to characterize particulars, such as *a lump of gold*.
FPSs, which involve an FU as well as finite verb, and perhaps some other parts, 'yield a stament of the incidence of the universal feature they introduce.'
Though FPSs do not require the introduction of particulars, the facts they state are presupposed by introductions of certain kinds of particular. An FPS such as 'There is water here' states a fact which is a precondition for the fact statable using, in part, 'This pool of water'.
This transition involves the addition, beyong the FU, and its simple language, apparatus for distinguishing and reidentifying particulars, as well as characterizing universals appropriate to that kind of particular.
He toys with the idea that a basis for reidentification may already exist at the FP level -- as one can say, 'There is snow here -- and here -- and here.' And thus a mutiplicity of placing may be a sort of proto-distinguishability. He drops this thought, with a promise to pick it up later.
Even though these feature-facts may be presupposed by the introductions of the sorts of particulars he has mentioned (pools of water, lumps of gold), this is not enough, as it should be the case that the facts presupposed2 by PI2 are such that they are P2ed by the introduction of all sorts of particulars. S reminds us that the argument of part one of the book was that a certain class of particulars -- directly locatable material bodies -- was basic to all particular identification. Thus, if we can find facts which are P2ed by introduction of these particulars, and which provide the conditions for their introduction, the general strategy will be vindicated.
The problem is that the particulars mentioned so far, lumps of gold and pools of water, seem to be atypical. Their names have a part, 'gold' or 'water' which is immediately amenable to an FU. But more typical sortals, such as 'men', 'cats', etc., do not obviously follow this model. It is difficult to imagine operating with FUs appropriate to cats and men, as if these were mass nouns, in short.
But this does not mean that it is impossible to imagine a level of thought at which such FUs are used (Langacker gives many examples of this -- after inadvertantly driving over his pet, he 'got cat all over his driveway', where 'cat' is a mass noun, which can be added to and subtracted from -- more cat (if he hits another), less cat (if he cleans some of it up). That is, Langacker also argues for particular identification as a cognitive construct from more basic ingredients (interestingly, for actions as well as things, but I'm getting too far afield).)
S introduced the notion of a naming game, which is just the reporting of the presence or absense of some feature (neutrally conceived) in the vicinity. E.g. 'Duck', 'ball', not to be conceived as 'a duck', or 'a ball'.
To the argument that 'cat' is unlike 'snow' in that it is not possible to coherently conceive of cat unless as tied to some particular cat, S replies, p. 207:
"What this argument shows, however, is not that the required type of general concept of cat is impossible, but rather that the concept must already include in itself the *basis* for the criteria of *distinctness* which we apply to particular cats. Roughly, the idea of the cat-feature, unlike that of snow, must include the idea of a characteristic shape, a characteristic pattern of occupation of space."
Likewise, the concept of a cat-feature provides the *basis for* (though it does not itself provide/impose) the criteria for reidentification, again, because the caracteristic spatial pattern of cat-instantiation allows for the notion of continuity through space-time of instances of that pattern.
p. 207:
"The decisive conceptual step to cat-particulars is taken when the case of 'more cat' or 'cat again' is subdivided into the case of 'another cat' and the case of 'the same cat again'."
S then dispenses with the idea that cat-feature really means something like cat-slice. But he does take the following seriously, which is that particulars like 'flash' and 'bang' do not seem to admit of a clear distinction between possession of the feature universal and possession of the idea of the particular. His answer is that these sorts of particulars in general are reindentified wrt locations and objects, and aren't that central to our conceptual scheme anyway.
I think he gives up too quickly on these. It seems to me that there is indeed a flash-feature, which is just 'light', exactly parallel to the fact that 'gold' is a feature of nuggets (i.e. lumps of gold). That is, things like bangs and beeps and flashes are just lexicalized versions of feature-cum-sortal expressions. A beep is a pure tone (feature) with very limited temporal boundaries, as a flash is brightness with limited temporal boundaries, and a nugget is gold with limited spatial boundaries. Langacker discusses such cases in 'nouns and verbs'.
7. The logical complexity of particulars and the 'completeness'
of logical subject-expressions. Particulars the paradigm logical
subjects.
S began with the notion that particulars had a certain kind of factual complexity which universals lacked. One way of resolving this complexity is as of a fact, expressed in a proposition, which as a whole individualtes the particular. Such propositions ultimately give way to others which introduce or quantify over particulars. But as this is a theory of PI1, this is not a defect.
The search for presuppositions2 leads to feature universals as expressing facts which underlie the application of particulars in statements in general. Since these underlie the introduction of basic particulars, they are presuppositions of introducing any sort of particular whatsoever. The particular thus 'unfolds' into a complete thought.
But the particular is also incomplete, because qua particular it is thought of as a constituent of some further fact, or thought.
Basic facts on this scheme, facts which do not themselves presuppose any particulars, are FP facts.
The rest is just summary of the chapter.
p. 212:
"To sum up, then. The aim has been to find a foundation
for the subject-predicate distinction in terms of some basic antithesis
between 'completeness' and 'incompleteness.' This antithesis was
to explain the traditional association of the subject-predicate
distinction with the particular-universal distinction. We find
this antithesis by, as it were, putting pressure on the idea of
a particular until it gives way to the idea of a fact. At the
limit of pressure we find the feature-placing fact in which no
particular is a constituent, though a universal is. At this limit,
then, the universal appears as still something incomplete for
thought, a constituent of a fact, whereas the particular does
not appear at all; and at this limit, we say, the antithesis,
subject-predicate, disappears. So we set up, as a paradigm for
reference, as a paradigm for the introduction of a subject, the
use of an expression to introduce a particular, to introduce,
that is, something which is both complete for thought in that
it unfolds into a fact, and incomplete in that, so introduced,
it is thought of as a constituent of a further fact; and we set
up as a paradigm of description, of the introduction of a predicate,
the use of an expression to introduce a universal, to introduce,
that is, something which has the same kind of incompletenesss
as the particular but lacks its completeness. The two introduced
terms are to be such that the assertion of a non-relational tie
between them constitutes something once more complete, a complete
thought; and the association of the symbolism of this assertion
with the universal rather than the particular we see, in the end,
as no more than a mark of the former's lack of completeness which
the latter possesses."