Thursday, November 16, 2017
Friday, April 28, 2017
BACKGROUND TO THE TRACTATUS (unedited notes)
BACKGROUND TO THE TRACTATUS
I. THE LINGUISTIC TURN: LOGICISM
1. KANT: Mathematics and the Synthetic A Priori
2. FREGE:Logicism and the foundations of Mathematics
a. The context principle
a. "How, then, are the numbers to be given to us, if we cannot have any ideas or
intuitions of them? Since it is only in the context of a sentence that words have
any meaning, our problem becomes this: To explain the sense of a sentence in
which a number word occurs." ((Frege, 1953), §62)
b. "...never ask for the meaning of a word in isolation, but only in the context of a
sentence." (p. x)
c. "It is enough if the sentence taken as a whole has sense; it is this that confers on
its parts also their content." (§60)
d. "We next laid down the fundamental principle that we must never try to explain
the meaning of a word in isolation, but only as it is used in the context of a
sentence." (§106)
The Context Principle in MT:
CP:
CP is used by Frege to defend two interrelated claims (Frege's theory of numbers)
FN:
(i) numbers are objects
(ii) numerals are genuine proper names
CP is used in anticipation of the following objection to FN:
OF:
(i) when we try to understand what a numeral names, we can form no idea/image of the meaning of numerals (self-evident thought experiment)
(ii) Therefore, numerals cannot really be names
Frege charges that OF is mistaken, because in order to understand the meaning of a numeral we do not need to be able to understand the meaning of a numeral in isolation from its role in propositions. CP says something stronger: it says, in its strongest form, words ONLY have meaning in the context of a proposition; and in a weaker "meta-semantic" form, to explain the meaning of a word requires understanding its role in propositions. There are issues with cohering these readings to Frege's formulations; also, issues with squaring it with Frege's further principle of compositionality. However, at any rate, for our purpose it is clear that in either reading it is suited to respond to OF: numerals can be understood in their propositional context, even if there are difficulties understanding their meanings in isolation.
[fill in the rest of Frege's strategy: how does the need to understand numerals within propositional context support or enable his definition of numerals in terms of equinumerous sets? Is the goal to 'recarve' numeral in terms that involves no reference to numbers or identity, since such definitions would be circular? How does Frege's concern with = lead to SR?
Potter in NL sees Frege's CP as an attack on psychologism, which presumably was a part of what resisted numerals as names claim:
"The moment Frege took the linguistic turn--the moment, indeed, when it was born--occurren in the Grundlagen when, having stated teh context principle, that 'only in the context of a proposition do words mean anything', he then used the principle to tranform the Kantian question, 'How are numbers given to us?' into the linguistic question, 'How are number words used?' Frege explicitly advertised one of the purposes of the context principle as being to resist psychologism. If we ignore this principle, he sia, we are 'almost forced to take as the meanings of words mental pictures or acts of the individual mind', and hence, he thought, to descend into psychologism." 65
Potter also claims that while for Frege the linguistic turn meant looking to language to approach thought, for Wittgenstein it was a symbolic turn which mean, looking to our symbols to approach the world.
Corollaries:
What follows from the claim that the smallest meaningful unit of language is the sentence? Nothing is said before the level of a sentence. This seems suited for a view of language that sees its essence in asserting truths: if the meaning of a linguistic entities is its relation to truth, then sentences are meaningful as such, while any part of a sentence is only meaningful insofar as it contributes to the truth conditions of the sentence.
This has prima facie appeal, but also is unintuitive. Let's consider the following: On the one had, in a dictionary, a words meaning is explicated by more words and phrases, sometimes simply synonyms. One the other, a good dictionary also shows the word in a sentence. At any rate, any dictionary will classify the word by its part of speech which determines the possibilities and impossibilities of the word's formal contribution to a sentence. Clearly, synonymy is not all that we are asking for when we ask for the meaning of a word. It is rather as though a dictionary gives the form and the content of a word - the content being given by more familiar words (something roughly like a possibility of substitution), the form being its part of speech - possiblity of meaningful combination. What makes it clear that even a dictionary is committed to some thing like the context principle is that by classifying a word by its part of speech is saying that its definition is incomplete without understanding what its part of speech is and moreover to understand parts of speech is essentially to understand something in terms of its ability to be part of a sentence. So in a minimal way, even a dictionary definition which seems to define a word in isolation still essentially involves seeing the isolated word as an abstraction from its possible combinations.
Still, there seems to be a difference between saying that one can't understand what an adverb without understanding a whole host of other parts of speech, not least of which is the sentence, and, on the other hand, saying that one can't understand what an adverb is outside of its actual occurrence in a sentence. I think what we need to posit is an ambiguity in Frege that amounts to a misleading interpretation of CP. To say, in the stronger reading of CP, "only in the context of a sentence does a word have any meaning" means: words do not have meaning until they actually occur in a sentence. This has the unintuitive conclusion that coming across a word on its own is meaningless. Certainly seeing the word "seashell" on a piece of paper is more meaningful than seeing the letters "seashe". In the same way something holistic occurs when a string of words becomes a sentence, so too something holistic emerges when a string of letters/phonemes amounts to a word.
On a weaker, more charitable reading, all Frege is saying by "context of a sentence" is something like: in its possible legitimate combinations; in its potential contribution to a whole sentence. Then the claim that words aren't meaningful on their own amounts to the claim that in order to understand any given isolated word, one must first understand it as an essentially incomplete entity - one which bears an essential but incomplete relation to truth (this is its meaning) but must combine with others to do anything properly meaningful.
It becomes important to posit something like a spectrum, on one end of which are names which refer to concrete observable objects; on the other, names that refer to abstract 'entities'. The former are paradigmatic names - things like "Ludwig Wittenstein" or "flour" on a jar. These names seem to be exhausted by brute correlation. Of course, this is not the case: the name sewn into a pair of coveralls is not the name of the coveralls, but the normal wearer, the "flour" on the jar refers to what is normally in the jar, even when its empty; however it is by trying unsuccessfully to imagine the abstract entities in terms of the paradigmatic cases that OF gets off the ground. However it quickly becomes clear that the very idea of a meaningful brute correlation between a word and an object is a fiction - it is impossible to understand "Flour" as a meaningful label without seeing it as shorthand for, or indicative of, certain claims of facts: ie., propositions (in this case, this jar is where the flour is kept; the intended wearer of these coveralls is called "Ludwid" etc.)
It turns out then that the CP amounts to a repudiation of the claim that language bottoms out in ostension: that, whether as a gesture or some simple, pointing relation between a label and its object, ostension is insufficient to account for the meaning of any bit of language. CP, then, is mustered to show that numerals can be understood as names, and the only reason this seems dubious is because we have something like an ostensive account of how names are meaningful, that the relation between a name and an objects is that the former simply stands for the latter, and nothing else. Here I will quote at length from MT:
"Since 'a word has meaning only in the context of a sentence', there must be a great deal more to fixing the meaning of a term than the bare assignment of referent. On the contrary, its 'logical grammar' must also be fixed - it must be determined how it will fit together with other expressions to form a significant sentence, what what basic principles of inference will govern the result, and so on. It is just not intelligible that all of this could be conveyed in a simple ostensive definition, or by any sort of pre-linguistic presentation of a referent. On the contrary, understanding an ostensive definition must presuppose considerable linguistic competence, since you will need to know what sort of thing is being pointed out to you, and the criterion of identity for things of that kind" MT 16
How does the rejection of the primacy of ostensive definition and the adoption of the context principle entail something like the "linguistic turn"? [This conclusion belongs to Carruthers, likely not to Frege]
A. In the weaker way, it is by retreating from claims about numbers to claims about statements about numbers, Frege pioneers a strategy, one that supposedly moves onto a more tractable terrain.
B. In the stronger way which results from the rejection of primacy of ostensive meaning; this seems to entail something like: we cannot settle what kinds of things there are in the world and how they are by some kind of pre-linguistic looking and seeing. The result is a kind of kantian-esque linguistic idealism: we cannot come to the world to inspect without presupposition what it contains, because in order to identify an object we must already understand it as classified and identifiable in such and such a way. The moral is not necessarily that we impose our grammar onto the structure of the world, but only that what we are really doing when we are going to investigate the world is investigating our own linguistic structure.
LT: Talking about the world (ontology) by Talking about a suitable (ideal or ordinary) language
(1) CP
(2) Therefore, the basic relation of language to the world cannot be one between isolated names and bare objects. (Failure of ostension)
(3) What we are investigating when we investigate the structure of the world is the structure of our language.
Back to MT:
"It is not just over the existence of the numbers, but in metaphyics generally, that Frege adopts this sort of approach. For example, he takes it as established that the entities which constitute the referents of predicative expressions must be in some way 'incomplete' because of his belief that such expressions themselves, and their senses, are essentially incomplete. So what emerges is that the context principle can be seen to underlie a great deal of Frege's approach to philosophy. Given the primacy of the sentence, and the impossibility of conferring meaning on individual terms by any sort of bare presentation of a referent - indeed since there can be no such thing as non-linguistic access to metaphysical truths - he thinks the only way of coming to discern the essential nature of reality is through the study of language"
Of course, this is a contentious characterization of Frege's understanding of his own project.
[Not saying: words are only meaningful in the context of a sentence (a semantic claim about the reference of words); rather he's saying: in order to explain the meaningfulness of words, they must be examined in the context of the sentences they meaningfully contribute to. His claim is not primarily a semantic one, but meta-semantic, or, one that claims about the proper methodology of semantics. It is difficult to understand formulations (a) and (c) as meta-semantic, however (b) and (c) do seem to be principles for determining referent, not claims about reference. Moreover, context-as-semantic can account for this difference (since context-as-semantic can entail context-as-metasemantic (it can even be explanatory of it)) while context-as-metasemantic cannot. One can claim that Frege's is being imprecise in cases (a) and (c).]
The linguistic turn means there is a distinction to be drawn between linguistics (which studies language qua language) and philosophy of language (which studies the nature of reality via language. What justifies the move from empirical, perhaps universal claims about language to transcendental or ontological claims? How do we make sense of taking ordinary language, with all its compromises based on its purpose of communication, and analysing it into an ideal language? Ideal for what? Not for communication...then for representation? For truth? For philosophy? What sense can we make of a linguistic ideal that is other than the common purpose of every natural language? The only correct answer would be ideal in its capacity to picture the world. However there is more to it than that - for although at work is some kind of assumption that language will reveal the world, or give us substantive non-linguistic answers, there is the contrary assumption that language distorts the world, as evidenced by Russell, Wittgenstein, Carnap, etc.
The linguistic turn then seems to be much more complex than simply letting language reveal the world: some times language is the indepenent variable and the world the dependent, but other times it is the world that is independent and language must be made to cohere - hence the claims of deep grammar, ideal languages etc. This indicates that it is never innocent language that is guiding the investigation, but always looking to language as a tool while trying to work out other commitments. Otherwise, how could the tractatus give us objects that correlate to names which can hardly be imagined (more like theoretical postulates) and names which are unlike any name anyone has ever used? The linguistic turn seems more about imposing a grammar on language given other theoretical commitments (numerals as names, making sense of empty reference), not about reading off of language what it 'speaks', but of injecting into language what it must be like if it is to be right (wrt one's logical apparatus, one's epistemic commitments etc, ones desire to eliminate philosophical problems in one fell swoop). In fact, I think it is impossible to divorce the claim of taking our lead from language from the results it had early on in solving philosophical problems; that is, I wish to distinguish between the claim that language reveals reality from the claim that language when properly analysed can dissolve philosophical problems. There are probably no philosophically compelling reasons to believe that the structures of language reveal the independent structures of the world (ie., the linguistic turn will be neutral wrt realism, idealism, skepticism) but there is evidence that logical analysis can dispel conflict. Certainly as much to the claim that language can reveal reality is the claim that language is misleading. What intuitions guide us, tell us, when it is doing which?
Philosophical questions are questions about language. At least it is clear that what is at stake here is also what is at stake when Frege questions whether identity statements are questions about language or their referents? However: Hacker in LT:
"Frege’s primary concern was to demonstrate that arithmetic is derivable from logic. It was tothat end that he invented his function-theoretic logic. He conceived of his logical system as an ideal language for logical and proof theoretic purposes. It was, he suggested, related to natural languages as the microscope to the eye. His philosophical attitude to natural language as a tool for the purposes of the philosophy of logic and mathematics was one of contempt. Natural languages did not evolve for the purposes of logical proofs; for that purpose one needs to invent a logically perfect language – which is what he presented his ‘conceptual notation’ as. This, broadly speaking, was also Russell’s view. He conceived of the Peano-derived symbolism and of the formation rules of Principia as the syntax of a logically ideal language."
Hacker claims that it is inappropriate to label Frege as the founder of the turn, since his ambitions for his language were limited to logical proofs, and since he:
"...had no general view of the sources of, nature of , or methods of solving, philosophical problems. He did not hold that all or even mostphilosophical questions are questions of language (which, according to Rorty, is one aspect of the linguistic turn). Nor did he claim that all or even most philosophical questions are to be answered or resolved by either examining the use of natural language or by inventing an ideal language (which, Rorty held, characterizes the linguistic turn. His concerns were exclusively with the philosophy of mathematics, logic and philosophical logic. And he invented his conceptual notation for purposes of his logicist project – not to solve or resolve the problems of epistemology, metaphysics, philosophy of mind, etc."
Frege's turn is better described as the logistic turn. It seems that he didn't work out all the implications of the context principle outlined above. "All philosophy is a critique of language" (Tr) This marks the beginning of the linguistic turn (it makes sense that Russell and Moore are still pre-linguistic turn thinkers. Compare Kant's critique of the ontological argument "Existence is not a perfection" versus a linguistic critique of it "existence is not a predicate".
The strongest claim to what language can reveal is not the nature of reality, but the limits of thought, of the thinkable. It seems compelling that if we cannot meaningfully say something then we cannot meaningfully think it.
b. The problem of identity (cognitive difference between same reference)
[3.] WITTGENSTEIN: the tractatus
[a.] The context principle
[b.] The problem of identity
II. THE OBJECTIVE TURN: IDEALISM, PSYCHOLOGISM; OBJECTIVITY AND ANALYSIS
1. LOCKE: Meaning as mental
2. BRADLEY: Idealism
a. Internal Relations
b. Partial Knowledge
c. Absolute Idealism and Skepticism
3. MOORE: Against private meaning
a. An equivocation in Bradley: IDEA as universal meaning, as occurent event
a. Proposition, concept, and objectivity
4. FREGE: ANTI-PSYCHOLOGISM
a. Logic: natural and normative laws
III. THE PROBLEM OF THE THINNESS OF OBJECTIVE MEANING
1. The problem of empty reference:
a. fictional entities (Pegasus), true ascriptions of non-existence (Santa doesn't exist)
b. non-existent entities
c. Single reference, different names, different cognitive content
2. MEINONG: Subsistence
3. FREGE: SENSE AND REFERENCE
4. RUSSELL: HIDDEN LOGICAL FORM SURFACE/DEPTH
IV. LANGUAGE AND PROPOSITIONAL UNITY: FREGE AND RUSSELL
a. FREGE's TWO STEP SEMANTICS: linguistic entity, sense, and referent
1. Name, mode of presentation and object
2. Predicates, relations, qualities and concept and unsaturated terms (truth functions)
3. Logical Operators and truth functions
4. Propositions and, thought and the True: Saturation and Propositional Unity
b. RUSSELL's ONE STEP SEMANTICS: linguistic entity, object of acquaintance
1. Name and sense data
2. Predicates, relations, qualities and formal acquaintance
3. Logical Operators
4. Proposition and Judgment: Multiple relation theory
V. WITTGENSTEIN
1. OUTSTANDING PROBLEMS FROM THE TRADITION
1. Simples: Name and object
a. 'necessary existence' and empty reference
2. Propositional unity: proposition and fact
3. Picture Theory of Meaning
a. Predication: correlation versus index
b. fact ontology
c. Truth and sense and nonsense
4. Saying and Showing: form and content
a. Logic and logical form
"The most influential achievement of the book was its clarification of the nature of logical truth. This was done by an investigation of symbolism. It was argued that the ‘peculiar mark of logical propositions [is] that one can recognize that they are true from the symbol alone, and this fact contains in itself the whole philosophy of logic’ (TLP 6.113). Contrary to what both Frege and Russell thought, the propositions of logic are not essentially general (but essentially true), they say nothing at all, but are rather senseless, i.e. limiting cases of propositions with a sense. In particular, they are not descriptions of relations between thoughts as Frege supposed, nor are they descriptions of the most general facts in the universe as Russell had suggested. LT
5. Philosophy
a. Science and natural law
b. Ethics/Aesthetics
c. Solipcism, realism, idealism
d. Metaphysics
6. The Paradox
a. The Traditional Reading
b. The Resolute Reading
c. Another Possibility?
7. MISCELLANEOUS
Saturday, April 22, 2017
Same Form
The sameness of form underpins the idea of the model theory of language. Something like: what can be described in language is the same as what can be the case in the world.
The problem is that the possible combinations of objects in the world seems obviously more restricted in their combination possibilities than language. To respond to this, W will ensure that the combination possibilities of language are pinned to the possibilities of the world. This will lead him to postulate a most restrictive account of possible linguistic combinations. What strikes me is that he seems to be engaging in the very method he proscribed for Russell, wrt theory of types: that one cannot dictate to grammar what it can do. [] Here I take the implication to be that grammar cannot be restricted by other, non-logical or non-descriptive commitments: One cannot, in order to achieve some non-descriptive goal, impose on language a logic of convenience, if you will. I wonder then, how much precisely this kind of trap is where the Tr ends up. What is the tension between a broad strategy begun by Russell: namely, the appeal to a hidden logical form that gets one out of certain complications, and which eventually becomes a generalized strategy (hidden existential quantifier: work of empty reference is done by an (unproblematically unfullfilled) variable, on one hand, AND a broadly descriptivist approach to what could count as good grammar.
Even if reality and the expressive possibilities of language end up being identical in a strong sense, what are we left with if it turns out that we can't really say anything about language? This it seems was an important advantage of the linguistic turn. What remains left of it in W's project, even if he does show argumentatively that there is such an identical underlying form? It may seem the very notion of a contentful remainder - what's left when we put empirical claims to one side and non-sense to the other - is inconsistent with an approach that took much as one of its guiding criteria the denial of the synthetic a priori.
The problem is that the possible combinations of objects in the world seems obviously more restricted in their combination possibilities than language. To respond to this, W will ensure that the combination possibilities of language are pinned to the possibilities of the world. This will lead him to postulate a most restrictive account of possible linguistic combinations. What strikes me is that he seems to be engaging in the very method he proscribed for Russell, wrt theory of types: that one cannot dictate to grammar what it can do. [] Here I take the implication to be that grammar cannot be restricted by other, non-logical or non-descriptive commitments: One cannot, in order to achieve some non-descriptive goal, impose on language a logic of convenience, if you will. I wonder then, how much precisely this kind of trap is where the Tr ends up. What is the tension between a broad strategy begun by Russell: namely, the appeal to a hidden logical form that gets one out of certain complications, and which eventually becomes a generalized strategy (hidden existential quantifier: work of empty reference is done by an (unproblematically unfullfilled) variable, on one hand, AND a broadly descriptivist approach to what could count as good grammar.
Even if reality and the expressive possibilities of language end up being identical in a strong sense, what are we left with if it turns out that we can't really say anything about language? This it seems was an important advantage of the linguistic turn. What remains left of it in W's project, even if he does show argumentatively that there is such an identical underlying form? It may seem the very notion of a contentful remainder - what's left when we put empirical claims to one side and non-sense to the other - is inconsistent with an approach that took much as one of its guiding criteria the denial of the synthetic a priori.
Wednesday, April 12, 2017
Wittgenstein: Unity of Proposition "Only facts can express a sense, a set of name cannot"
As a result of an attempt to account for the unity of the proposition.
So:
3.14 What constitutes a propositional sign is that in its elements (the words) stand in a determinate relation to one another.
A propositional sign is a fact.
[What's of first importance here is that the essence of a sentence is it's being a fact. This should be differentiated from the fact that the sentence expresses. It can only express a fact about the world because it shows a fact about itself: namely, it is a determinate relation of linguistic elements.
Thus, a proposition is not just a "blend of words" nor a "set of names". In 3.143, Wittgenstein, in a criticism of F & R, explains how one might take a propositional sign as a thing, or as namelike. "in a printed proposition, for example, no essential difference is apparent between a propositional sign and a word." He is saying, perhaps then, that just as words are made up of letters, so one might model the understanding of sentences on words, but now composed of words in turn. What we seem to be left with (as in Frege's case) is a more complex name - a "composite name".
Maybe then W's claim is that this way of going astray leaves one with an account of the sentence, the "propositional sign", that really amounts to little more than a "blend of words" or "set of names". This could also be what led Frege to resort to his assertion quantifier, which made propositions more like 'assumptions' that might, as a matter of fact, have something like assertion tacked on.
He elaborates on this by suggesting we imagine a propositional sign composed of objects rather than letters and words. So while the sentence "Socrates is bald" as a whole may just seem to bear the same relation to its meaning as the parts do, it is harder to make this mistake if we imagine, as W did, the kind of proposition that is asserted by say a courtroom car-accident model: here we are less tempted to see the arrangement of the model of say two autos as some third thing on ontological par with the autos. Rather, the model example shows more clearly that the power of the propositional sign is that it itself is a state of affairs, ie, it itself is a fact: the fact that the two model-autos are thus arranged means that car-a rear ended car b. It is a fact about a propositional sign that expresses a fact about the world.
This is the meaning of 3.1432:
Instead of, [1] 'The complex sign 'aRb' says that a stands to b in the relation R',
we ought to put,
[2] 'That "a" stands to "b" in a certain relation says that aRb.'
Here, [1] is trying to analyze the propositional sign "aRB" in terms of three elements: a, b and relation-R, where the relation of "a" to a and "b" to b is taken to be the same as the relation of "R" to r. Thus we have a set of names.
But in [2] we're trying something different. We're not referring to the relation-R as a referent of R, rather, we're talking about a fact - "a" standing in a certain relation to "b" which does not try to refer to R. Rather, the idea is elucidated by the courtroom model - there is not the two autos AND there relation-R which are elements of the model's ontology. Rather, properly speaking, only modelcar-a, modelcar-b correlate with any things in the world; there relation - which unproblematically has no correlate in the model or indeed anywhere - is a fact that can be seen by looking at them, just as the space and letter "R" establishes a relation between two letters, "a" and "b". This relation may seem to be meaningful in the same way that "a" and "b" are, but it is not. We might say what is meaningful in the proposition "aRb" is the fact that "a" and "b" have an "R" between them (i.e., they are related in a certain way, just as the modelcars were related in a certain way in the model.)
What is confusing is that relation terms superficially look just like names. Plato teaches Aristotle. The courtroom model (which is probably not just a propositional sign analogy but just IS a propositional sign, and is also probably why W speaks of prop signs rather than sentences) shows however that it is by virtue a convention of relating the linguistic elements (which get their meaning correlatively, ie., names or model-cars) in differentiatable way (it need not be through another word, as the example of an exponential function shows).
The question is of course, how do we know which Fact is picked out by the fact that "a" stands in some kind of recognizable way to "b"? Don't we still need to lean on the idea of correlate relations so that the right facts are depicted by the right propositional signs? I'm not sure I understand the solution proposed, but here is Potter:
"What is expressive in not the complex consisting of three signs, but a fact about this complex, namely, that in it the sign 'R' occurs with something to the left of it and something else to the right of it. The sign 'R' functions only as a label to distinguish this relationship between the names 'a' and 'b' from other possible relationships (my emph) (such as the one exemplified in 'aLb', for instance). ...let us call signs used as labels in this fashion indices. The correlation between an index and what it labels is a matter of convention. Indeed, the fact that there is an index at all is conventional. What is required in order to express that aRb is only that the names 'a' and 'b' should stand in a certain relationship. In the present case, of course, the relationship is, as it happens, that of having the index "R" standing between them, but there is nothing essential about this. It would be perfectly possible to have a language in which some relations are expressed not with the aid of indices but rather by other devices such as spatial relations between the signs." 114
I'm just not sure why this "correlation" between arbitrary assignments of symbolic relations and all other expressible relations isn't problematic in the same way that e.g., the relation between names and objects is problematic (look at where it leads R and W) or in the same way all the issues about abstract objects and universals etc are problematic. The courtroom model can unproblematically model the car crash for two reasons: (1) they are both using objects that are related in the same medium - i.e., space; and (2) we possess a scale for translating the space of one to the model space of the other. However the precisely difficult thing about linguistic referent is that it does not for the most part share its differentiating structure with the things its talking about. The sense making of the model would have to be much more difficult than the courtroom case. However, W is not at this stage interested in these kind of 'practical' details, if indeed his claim grounds somethings possibility, the details are not important. OR what is more unsettling is that there is really only one kind of relation between simple objects, viz., spatial, in which case, trying to understand the complex correlations between all the linguistic relations and their extra-linguistic counterparts would be as fruitless, and for the exact same reasons, as trying to understand what a simple objects are like. It only matters that at some, unimaginable level, such objects and such an index is possible.
This contributes to the fact-ontology:
"In order to make judgments about the world, what we must perceive are facts, not complexes; and the symbols that express those judgments, likewise, must be facts, not complexes". 114
So:
3.14 What constitutes a propositional sign is that in its elements (the words) stand in a determinate relation to one another.
A propositional sign is a fact.
[What's of first importance here is that the essence of a sentence is it's being a fact. This should be differentiated from the fact that the sentence expresses. It can only express a fact about the world because it shows a fact about itself: namely, it is a determinate relation of linguistic elements.
Thus, a proposition is not just a "blend of words" nor a "set of names". In 3.143, Wittgenstein, in a criticism of F & R, explains how one might take a propositional sign as a thing, or as namelike. "in a printed proposition, for example, no essential difference is apparent between a propositional sign and a word." He is saying, perhaps then, that just as words are made up of letters, so one might model the understanding of sentences on words, but now composed of words in turn. What we seem to be left with (as in Frege's case) is a more complex name - a "composite name".
Maybe then W's claim is that this way of going astray leaves one with an account of the sentence, the "propositional sign", that really amounts to little more than a "blend of words" or "set of names". This could also be what led Frege to resort to his assertion quantifier, which made propositions more like 'assumptions' that might, as a matter of fact, have something like assertion tacked on.
He elaborates on this by suggesting we imagine a propositional sign composed of objects rather than letters and words. So while the sentence "Socrates is bald" as a whole may just seem to bear the same relation to its meaning as the parts do, it is harder to make this mistake if we imagine, as W did, the kind of proposition that is asserted by say a courtroom car-accident model: here we are less tempted to see the arrangement of the model of say two autos as some third thing on ontological par with the autos. Rather, the model example shows more clearly that the power of the propositional sign is that it itself is a state of affairs, ie, it itself is a fact: the fact that the two model-autos are thus arranged means that car-a rear ended car b. It is a fact about a propositional sign that expresses a fact about the world.
This is the meaning of 3.1432:
Instead of, [1] 'The complex sign 'aRb' says that a stands to b in the relation R',
we ought to put,
[2] 'That "a" stands to "b" in a certain relation says that aRb.'
Here, [1] is trying to analyze the propositional sign "aRB" in terms of three elements: a, b and relation-R, where the relation of "a" to a and "b" to b is taken to be the same as the relation of "R" to r. Thus we have a set of names.
But in [2] we're trying something different. We're not referring to the relation-R as a referent of R, rather, we're talking about a fact - "a" standing in a certain relation to "b" which does not try to refer to R. Rather, the idea is elucidated by the courtroom model - there is not the two autos AND there relation-R which are elements of the model's ontology. Rather, properly speaking, only modelcar-a, modelcar-b correlate with any things in the world; there relation - which unproblematically has no correlate in the model or indeed anywhere - is a fact that can be seen by looking at them, just as the space and letter "R" establishes a relation between two letters, "a" and "b". This relation may seem to be meaningful in the same way that "a" and "b" are, but it is not. We might say what is meaningful in the proposition "aRb" is the fact that "a" and "b" have an "R" between them (i.e., they are related in a certain way, just as the modelcars were related in a certain way in the model.)
What is confusing is that relation terms superficially look just like names. Plato teaches Aristotle. The courtroom model (which is probably not just a propositional sign analogy but just IS a propositional sign, and is also probably why W speaks of prop signs rather than sentences) shows however that it is by virtue a convention of relating the linguistic elements (which get their meaning correlatively, ie., names or model-cars) in differentiatable way (it need not be through another word, as the example of an exponential function shows).
The question is of course, how do we know which Fact is picked out by the fact that "a" stands in some kind of recognizable way to "b"? Don't we still need to lean on the idea of correlate relations so that the right facts are depicted by the right propositional signs? I'm not sure I understand the solution proposed, but here is Potter:
"What is expressive in not the complex consisting of three signs, but a fact about this complex, namely, that in it the sign 'R' occurs with something to the left of it and something else to the right of it. The sign 'R' functions only as a label to distinguish this relationship between the names 'a' and 'b' from other possible relationships (my emph) (such as the one exemplified in 'aLb', for instance). ...let us call signs used as labels in this fashion indices. The correlation between an index and what it labels is a matter of convention. Indeed, the fact that there is an index at all is conventional. What is required in order to express that aRb is only that the names 'a' and 'b' should stand in a certain relationship. In the present case, of course, the relationship is, as it happens, that of having the index "R" standing between them, but there is nothing essential about this. It would be perfectly possible to have a language in which some relations are expressed not with the aid of indices but rather by other devices such as spatial relations between the signs." 114
I'm just not sure why this "correlation" between arbitrary assignments of symbolic relations and all other expressible relations isn't problematic in the same way that e.g., the relation between names and objects is problematic (look at where it leads R and W) or in the same way all the issues about abstract objects and universals etc are problematic. The courtroom model can unproblematically model the car crash for two reasons: (1) they are both using objects that are related in the same medium - i.e., space; and (2) we possess a scale for translating the space of one to the model space of the other. However the precisely difficult thing about linguistic referent is that it does not for the most part share its differentiating structure with the things its talking about. The sense making of the model would have to be much more difficult than the courtroom case. However, W is not at this stage interested in these kind of 'practical' details, if indeed his claim grounds somethings possibility, the details are not important. OR what is more unsettling is that there is really only one kind of relation between simple objects, viz., spatial, in which case, trying to understand the complex correlations between all the linguistic relations and their extra-linguistic counterparts would be as fruitless, and for the exact same reasons, as trying to understand what a simple objects are like. It only matters that at some, unimaginable level, such objects and such an index is possible.
This contributes to the fact-ontology:
"In order to make judgments about the world, what we must perceive are facts, not complexes; and the symbols that express those judgments, likewise, must be facts, not complexes". 114
Saturday, April 8, 2017
Russell
Russell - Hidden Logical Form
Project:
Russell's brand of Logicism led him to introduce the idea of a denoting term as an 'aboutness' shifter; he found this concept problematic, esp wrt talking about the meaning of such a term. This is raised the in Gray's Elegy argument. I need to understand this. Instead, he discovered a method for reconstruing sentences so to avoid reference to any entities that don't or may not have an objective correlate.
LING EXPRESS EX-LING CORRELATE
PART: word: term
- names -objects in the world (sense-datum)
- predicates -qualities, relations
WHOLE: sentence proposition
This is referred to as a one-step semantic theory. I'm not sure, beyond purely methodological considerations, why Russell doesn't take to sense. He is unconvinced that a proper name will need anything like a sense, though he of course sees the extra information asserted in a definite description. It might be simply that he thought his method of hidden logical form - i.e., hidden existential quantification, that we're speaking about some unique thing, and saying it has some properties - was sufficiently strong enough to deal with any cases of VAC, seen or unforeseen. He clearly sees some general application of this method, since he also tries to apply it to address the problem of unity of identity, also construing judgments as asserting the existence of a complex etc. At any rate, not only can empty definite descriptions be accounted for, but Russell feels entitled now to claim that anything that risks empty referent is actually just a disguised definite description: so "Phosphorus is Hesperus" really doesn't refer to any objects at all, but says states that some thing exists with unique properties, and some thing else does, and they are the same some thing.
This outlines a broad strategy: Russell now has found a way to deny that any given expression has correlative meaning, ie., names something in the world. Or: Russell only needs to admit correlative meaning to those linguistic entities as allowed by his other theoretical commitments.
Russell ends up with sense-data as the correlate of names because it is the only thing left over when everything that has the property of non-existence is shown to be a disguised definite description, but ALSO anything that coherently can be said to possibly not exist. We are not just concerned about the meaning of phrases with as a matter of fact non-existence entities in them, but also all coherent statements that something might not exist. We are not just plagued by the concept of actual non-existence, but also potential non-existence. If I can wonder whether "Plato exists", then Plato must be a disguised definite description. It is easy to see how only sense-data remain. This trajectory should've indicated that something had gone wrong, however Cartesian style world-regaining was more acceptable then - Husserl as well.
Project:
Russell's brand of Logicism led him to introduce the idea of a denoting term as an 'aboutness' shifter; he found this concept problematic, esp wrt talking about the meaning of such a term. This is raised the in Gray's Elegy argument. I need to understand this. Instead, he discovered a method for reconstruing sentences so to avoid reference to any entities that don't or may not have an objective correlate.
LING EXPRESS EX-LING CORRELATE
PART: word: term
- names -objects in the world (sense-datum)
- predicates -qualities, relations
WHOLE: sentence proposition
This is referred to as a one-step semantic theory. I'm not sure, beyond purely methodological considerations, why Russell doesn't take to sense. He is unconvinced that a proper name will need anything like a sense, though he of course sees the extra information asserted in a definite description. It might be simply that he thought his method of hidden logical form - i.e., hidden existential quantification, that we're speaking about some unique thing, and saying it has some properties - was sufficiently strong enough to deal with any cases of VAC, seen or unforeseen. He clearly sees some general application of this method, since he also tries to apply it to address the problem of unity of identity, also construing judgments as asserting the existence of a complex etc. At any rate, not only can empty definite descriptions be accounted for, but Russell feels entitled now to claim that anything that risks empty referent is actually just a disguised definite description: so "Phosphorus is Hesperus" really doesn't refer to any objects at all, but says states that some thing exists with unique properties, and some thing else does, and they are the same some thing.
This outlines a broad strategy: Russell now has found a way to deny that any given expression has correlative meaning, ie., names something in the world. Or: Russell only needs to admit correlative meaning to those linguistic entities as allowed by his other theoretical commitments.
Russell ends up with sense-data as the correlate of names because it is the only thing left over when everything that has the property of non-existence is shown to be a disguised definite description, but ALSO anything that coherently can be said to possibly not exist. We are not just concerned about the meaning of phrases with as a matter of fact non-existence entities in them, but also all coherent statements that something might not exist. We are not just plagued by the concept of actual non-existence, but also potential non-existence. If I can wonder whether "Plato exists", then Plato must be a disguised definite description. It is easy to see how only sense-data remain. This trajectory should've indicated that something had gone wrong, however Cartesian style world-regaining was more acceptable then - Husserl as well.
Frege
FREGE - SENSE MEDIATION
Project:
LOGICISM: Arithmetic can be redefined in strictly logical terms. Axiom regarding predicates definable in terms of class proves to generate a contradictory class.
PROBLEM OF IDENTITY:
LING EXP SENSE REFERENT
Singular terms (least restrictive) sense Objects
Sentences thought The True/the False
Predicates concept Truth Function
Logical Constant ? Truth Functions
SOLUTION TO VACUITY:
Sense: described as the manner in which a referent is picked out. Public, objective.
Introduced to solve VAC: introduces source of meaningful difference in cases of non-trivial identity. Does it explain MER? Are the meaning of sentences containing empty referents explained by having a sense but no referent? IF we allow that terms may be meaningful by virtue of having a sense but no reference, then does reference fall out of the picture? I'm sure this is covered somewhere.
LEAVES UNITY UNACCOUNTED FOR:
Frege seems to provide a solution to the unity of the sentence grounded in an objective, non-actual entity: false sentences have such a referent - the false. However it is unclear how explanatory that this can be, since all false (or true) sentences end up with the same reference. His account of thoughts will have to do the work here instead.
Project:
LOGICISM: Arithmetic can be redefined in strictly logical terms. Axiom regarding predicates definable in terms of class proves to generate a contradictory class.
PROBLEM OF IDENTITY:
LING EXP SENSE REFERENT
Singular terms (least restrictive) sense Objects
Sentences thought The True/the False
Predicates concept Truth Function
Logical Constant ? Truth Functions
SOLUTION TO VACUITY:
Sense: described as the manner in which a referent is picked out. Public, objective.
Introduced to solve VAC: introduces source of meaningful difference in cases of non-trivial identity. Does it explain MER? Are the meaning of sentences containing empty referents explained by having a sense but no referent? IF we allow that terms may be meaningful by virtue of having a sense but no reference, then does reference fall out of the picture? I'm sure this is covered somewhere.
LEAVES UNITY UNACCOUNTED FOR:
Frege seems to provide a solution to the unity of the sentence grounded in an objective, non-actual entity: false sentences have such a referent - the false. However it is unclear how explanatory that this can be, since all false (or true) sentences end up with the same reference. His account of thoughts will have to do the work here instead.
Friday, April 7, 2017
Historical Pressures
I. COMMITMENTS AND CONTENTIONS
COMMON COMMITMENTS:
1. (COR) CORRELATIVE MEANING: a linguistic entity is said to have correlative meaning (if and?) only if they are meaningful due to being correlated or indexed with an extra-linguistic entity.
- The correlation relation possesses (at least) these two features:
(ARB) arbitrary (external) relation
(NTR) non-transformative relation (this will certainly be required if mirror analogies are to hold)
This assumption is supposed to be an intuitive starting place. I take it that there is a further assumption that in the absence of any additional claims of how ling.ent get their meaning, that ALL linguistic meaning is, immediately or derivatively, given to linguistic entities in this way.
[As far as I can tell, this assumption is shared by all relevant parties wrt the Tr. It has been suggested that Davidson's semantics would present a denial and alternative to correlative meaning. This fits with what little I know about his epistemology, something to the extent that nothing extra-linguistic could give meaning to something linguistic, etc. Follow up.]
2. (OBJ) OBJECTIVITY: the extra-linguistic entities with which ling.ents are correlated are objects in the world.
- To be an object in the world is (at least) to be:
(PUB) publicly accessible
(IND) possess existence that is not dependent on being perceived or represented by anyone
[This is a common assumption to all relevant parties, and in fact this commitment seems to me as THE motivating presupposition/conclusion that founds the distinctly analytic movement, as central to understanding its unique problematic as Frege and Russell's logicism.
MOORE: Follow up: Bradley's idealism is to be the most significant statement of what this movement is a turn away from, with OBJ as the pivot. Bradley is taken to have been committed to the following claims
(i) that all relations are internal;
(ii) relatedly, that truth is always a matter of degree, and always incomplete;
(iii) some form of global skepticism, perhaps
Also, he is charged by Moore with a equivocal notion of 'idea'. Moore can disambiguate two usages (The nature of judgment) idea as an occurrent mental event and idea as universal meaning. From this latter he develops the notion of the concept, which is committed to (IND) and also the denial that a concept is a mere 'abstraction' from ideas or things: each of these is itself comes under the concept of concept as anything else (not sure if I get this or what work it's doing).
FREGE: Follow up (The thought) Frege's motives seem to come out of his Logicist project. This needs to be expanded on greatly; he has arguments for rejecting psychologism: charging that ideas as private mental acts fall out of the picture in an account of public meaning (is this anachronistic?); he also finds a conflation between descriptive or 'natural' laws, and prescriptive or normative laws. But the logicist project as such seems to demand a more 'substantial' status to the claims of logic if founding mathematics in logic is supposed to count as an accomplishment, or as a founding at all. Bringing in this dimension, we see also another target of criticism: Kant. In Begriffschrift, Frege declares he is searching for a system that will "bring forth judgments that at first sight appear to be possible only on the basis of some intuition." There is a broad project then that might be called (EMP) the empiricist project: to do away with the notion of the synthetic a priori, to explain how we can have knowledge of logic, metaphysics and mathematics without some kind of intuition. It is probably too broad reaching: I'm not sure Russell could be counted as not wanting to appeal to some kind of intuition, since his acquaintance seems to have, at least at times, applied not only to empirical entities but also things like universals and logical constants. Moreover, I get the impression from what little I've read that Russell is something of a one trick pony: beyond his treatment of definite descriptions, it's clear that his thought is more instructive as an exhibition of error than a font of philosophical methodological and substantive inspiration. Still, historically speaking, his often evolving attempts to solve the problems raised by his own current set of philosophical hypotheses is not simply explanatory of W's framework, but is telling to exhibit - perhaps by virtue of the strained positions he finds himself in - of the inherent difficulty of the problematic of early analytic philosophy and it's commitments to COR, OBJ, and perhaps EMP.
LOCKE: Follow up: Lockean phil of language founds the subjective explanation of linguistic correlative meaning. Locke seemed to believe that words most immediately refer to occurrent ideas in the speaker's mind. This may seem, for reasons analytic and continental, laughably primitive, however I do hope to show that the (early) analytic tradition, in many ways is dealing with problems that were easy enough to solve within a subjectivist or mental meaning framework. Most prominently, the problem of meaningful but non-existent entities or sentences seem simple enough for Locke to explain, however explaining the content of occurrent ideas might prove to force him into the same problematic.
(RFL) LINGUISTIC REFLECTION: structural properties of linguistic entities reflect/reveal structural properties of the world (or of the objective, extra-linguistic entities to which they correspond)
From this follows the crucial claim that to understand the former is to understand the latter, the former not being intrinsically interesting, or, at least, the understanding of the world as the more ultimate philosophical goal.
A crude an incomplete construction of the reasoning is something like:
(1) COR
(2) OBJ
(TF) RFL
I can understand the broad idea of this argument, but it seems that there are some crucial and contentious that allow the move from (1) & (2) to RFL. Some assumptions would be: a language will contain at least all the vocabulary and (more importantly) the grammatical forms to represent and so articulate the real categories of the world. The counter claim however seems imaginable: the language gets its meaning from its real world correlates, however it does not, even at some 'deep' level, possess the kind of fine-tuned sensitivity to capture all real categories. Of course, the weaker claim might just be that an analysis of language can, at best, reveal all the categorical distinctions that we can speak about. The ambition of these two different claims seems to be infinitely different to me, even if the results would amount to the same thing. I need to understand the justification for this move much better. So we can maybe see how eg post-structuralist thinkers reject RFL because they, at least, reject NTR condition of OBJ. If we, somehow, hold a view of language by which reality is transformed, constructed, or constituted by contingent linguistic frameworks, then it is easy to imagine the argument failing for want of a complete set of true premises. However it still seems questionable whether COR and OBJ alone entail RFL, whether the inference is valid. I suspect this question will come up in my prospective reading, and if not there, I doubt there is a shortage on the subject.
OPEN ISSUES:
(VAC) PROBLEM OF VACUITY - General term for problems raised by the move toward OBJ. There are at least two:
(MER) the problem of making sense of sentences which do not have an objective correlate e.g., "Superman is weak to kryptonite" or "The king of France is bald." This will be the first problem raised by the move to meaning grounded in objective correlation: how do we explain the way we understand the meaning of linguistic entities that do not have any such correlate.
MER seems to arise pretty immediately in response to COR & OBJ. It also is related to an old problem, already raised by Plato (in the Theatetus?). Competing strategies (positing another extra-linguistic layer (Frege), (analyzing sentence to expose logical form that makes no such reference, i.e., one that relies on objectively correlative meaning (Russell) will arise in line with broader theoretical commitments (Frege's light of making sense of communication, Russell's in light of epistemic considerations).
(NTI) the problem of making sense of non-trivial identities e.g., "Hesperus is phosphorus" or "2+3=5". Why should these sentence be meaningfully different from the trivial identity claims "Phosphorus is phosphorus" and "5=5", if the entities on both sides of the equation are meaningful in virtue of the same objective correlates.
NTI was raised by Frege as a way of defending his logicist project, needing to clarify a universal sense of identity (follow up: Elucidations).
(UNT) ACCOUNT OF UNITY - Any theory of meaning will have to explain the holistic unity of the sentence. This can be exhibited by considering the difference between a list and a sentence: you can remove and add items from a list and it remains a list, however removing a term from sentence may cause it to cease to be a sentence. Since linguistic terms are as such ARB (there is nothing about linguistic entities as such that ensures unity), there must be some extra-linguistic ground for this unity.
Propositions "cannot consist of names alone; they cannot be classes of names"...For a class of names offers no prospect of supplying the resources to explain the unity of the proposition" NL 111.
Accounting for unity without supposing OBJ seems more easily accomplished: Locke, and Kant?,for instance, held that it was the mental act of judgment that 'holds together' the elements of sentence: unity as a mental, occurrent act. But since OBJ is not supposing linguistic meaning to depend on mental acts, it seems it is committed to the idea that their is an extra-linguistic objective unity in the world that grounds the unity of the sentence. This will also be related to Frege's context principle, I think. Wittgenstein takes it as a given that one cannot 'judge a nonsense', so if judgment constitutes the unity of the sentence, then nothing would rule out judging a nonsense.
If we observe that the (SPR) meaningfulness of a proposition cannot depend on its truth, otherwise we could not make sense of falsehood, then the problem of unity can be seen as the problem of accounting for unity as a tension between the requirement for objectivity (that the extra-linguistic unity must be objective) and that false sentences must equally be possessed of that same unity: so we require an account of unity that is somehow objective but also non-actual. Russell, who is most stringently attached to the actuality (to the point of indubitability) of any objective correlates will struggle most to make sense of this. Frege seems to depend on making sense of the unity of thought. While W will deny that a sentence has an extra-linguistic entity from which it gets its meaning.
F,R and W all see that they must account somehow for the form of the proposition, that it is not the heap of names or terms that can account for the fact that something is asserted, but essentially an orderly combining of them. (a cobb salad is not just all the ingredients of a cobb salad). Early, Russel inspired solution involved casting a proposition as an existential assertion about a complex. Frege used the notion of function but this does not help with understanding a proper analysis of the proposition (scrutiny of a proposition cannot reveal the function whose value for the argument it is.)
Here Wittgenstein must add the notion of form to the picture, which must be something radically different from a name.
(GDV) QUESTION GRAMMATICAL DIVERSITY - This is a question about whether, or how many, logical/grammatical types their are. Possibilities will be horizontal (names, predicates) and vertical (as in a hierarchy of types) and unclear (logical constants).
The main force of the argument for an undifferentiated horizontal grammar is made by Russell. He seems above all committed to the idea that any term must be able to be spoken of as subject and as quality. It is easy to see the complications involved of differentiating or defining the category of predicate without making them objects and thereby referring to something essentially difference than what we want to be talking about. Ordinary language seems to allow a single concept to change its form as needed - a throw, throwing, thrown, etc. Intuitively there should be a single concept that these make reference to, however different grammatical types will make it not only impossible for the surface connection to be made between a throw and throwing, but it will make it impossible to differentiate them at the grammatical level. Frege also considers this problem, where he needs to say something like, "...is a horse" refers to the concept horse.
COMMON COMMITMENTS:
1. (COR) CORRELATIVE MEANING: a linguistic entity is said to have correlative meaning (if and?) only if they are meaningful due to being correlated or indexed with an extra-linguistic entity.
- The correlation relation possesses (at least) these two features:
(ARB) arbitrary (external) relation
(NTR) non-transformative relation (this will certainly be required if mirror analogies are to hold)
This assumption is supposed to be an intuitive starting place. I take it that there is a further assumption that in the absence of any additional claims of how ling.ent get their meaning, that ALL linguistic meaning is, immediately or derivatively, given to linguistic entities in this way.
[As far as I can tell, this assumption is shared by all relevant parties wrt the Tr. It has been suggested that Davidson's semantics would present a denial and alternative to correlative meaning. This fits with what little I know about his epistemology, something to the extent that nothing extra-linguistic could give meaning to something linguistic, etc. Follow up.]
2. (OBJ) OBJECTIVITY: the extra-linguistic entities with which ling.ents are correlated are objects in the world.
- To be an object in the world is (at least) to be:
(PUB) publicly accessible
(IND) possess existence that is not dependent on being perceived or represented by anyone
[This is a common assumption to all relevant parties, and in fact this commitment seems to me as THE motivating presupposition/conclusion that founds the distinctly analytic movement, as central to understanding its unique problematic as Frege and Russell's logicism.
MOORE: Follow up: Bradley's idealism is to be the most significant statement of what this movement is a turn away from, with OBJ as the pivot. Bradley is taken to have been committed to the following claims
(i) that all relations are internal;
(ii) relatedly, that truth is always a matter of degree, and always incomplete;
(iii) some form of global skepticism, perhaps
Also, he is charged by Moore with a equivocal notion of 'idea'. Moore can disambiguate two usages (The nature of judgment) idea as an occurrent mental event and idea as universal meaning. From this latter he develops the notion of the concept, which is committed to (IND) and also the denial that a concept is a mere 'abstraction' from ideas or things: each of these is itself comes under the concept of concept as anything else (not sure if I get this or what work it's doing).
FREGE: Follow up (The thought) Frege's motives seem to come out of his Logicist project. This needs to be expanded on greatly; he has arguments for rejecting psychologism: charging that ideas as private mental acts fall out of the picture in an account of public meaning (is this anachronistic?); he also finds a conflation between descriptive or 'natural' laws, and prescriptive or normative laws. But the logicist project as such seems to demand a more 'substantial' status to the claims of logic if founding mathematics in logic is supposed to count as an accomplishment, or as a founding at all. Bringing in this dimension, we see also another target of criticism: Kant. In Begriffschrift, Frege declares he is searching for a system that will "bring forth judgments that at first sight appear to be possible only on the basis of some intuition." There is a broad project then that might be called (EMP) the empiricist project: to do away with the notion of the synthetic a priori, to explain how we can have knowledge of logic, metaphysics and mathematics without some kind of intuition. It is probably too broad reaching: I'm not sure Russell could be counted as not wanting to appeal to some kind of intuition, since his acquaintance seems to have, at least at times, applied not only to empirical entities but also things like universals and logical constants. Moreover, I get the impression from what little I've read that Russell is something of a one trick pony: beyond his treatment of definite descriptions, it's clear that his thought is more instructive as an exhibition of error than a font of philosophical methodological and substantive inspiration. Still, historically speaking, his often evolving attempts to solve the problems raised by his own current set of philosophical hypotheses is not simply explanatory of W's framework, but is telling to exhibit - perhaps by virtue of the strained positions he finds himself in - of the inherent difficulty of the problematic of early analytic philosophy and it's commitments to COR, OBJ, and perhaps EMP.
LOCKE: Follow up: Lockean phil of language founds the subjective explanation of linguistic correlative meaning. Locke seemed to believe that words most immediately refer to occurrent ideas in the speaker's mind. This may seem, for reasons analytic and continental, laughably primitive, however I do hope to show that the (early) analytic tradition, in many ways is dealing with problems that were easy enough to solve within a subjectivist or mental meaning framework. Most prominently, the problem of meaningful but non-existent entities or sentences seem simple enough for Locke to explain, however explaining the content of occurrent ideas might prove to force him into the same problematic.
(RFL) LINGUISTIC REFLECTION: structural properties of linguistic entities reflect/reveal structural properties of the world (or of the objective, extra-linguistic entities to which they correspond)
From this follows the crucial claim that to understand the former is to understand the latter, the former not being intrinsically interesting, or, at least, the understanding of the world as the more ultimate philosophical goal.
A crude an incomplete construction of the reasoning is something like:
(1) COR
(2) OBJ
(TF) RFL
I can understand the broad idea of this argument, but it seems that there are some crucial and contentious that allow the move from (1) & (2) to RFL. Some assumptions would be: a language will contain at least all the vocabulary and (more importantly) the grammatical forms to represent and so articulate the real categories of the world. The counter claim however seems imaginable: the language gets its meaning from its real world correlates, however it does not, even at some 'deep' level, possess the kind of fine-tuned sensitivity to capture all real categories. Of course, the weaker claim might just be that an analysis of language can, at best, reveal all the categorical distinctions that we can speak about. The ambition of these two different claims seems to be infinitely different to me, even if the results would amount to the same thing. I need to understand the justification for this move much better. So we can maybe see how eg post-structuralist thinkers reject RFL because they, at least, reject NTR condition of OBJ. If we, somehow, hold a view of language by which reality is transformed, constructed, or constituted by contingent linguistic frameworks, then it is easy to imagine the argument failing for want of a complete set of true premises. However it still seems questionable whether COR and OBJ alone entail RFL, whether the inference is valid. I suspect this question will come up in my prospective reading, and if not there, I doubt there is a shortage on the subject.
OPEN ISSUES:
(VAC) PROBLEM OF VACUITY - General term for problems raised by the move toward OBJ. There are at least two:
(MER) the problem of making sense of sentences which do not have an objective correlate e.g., "Superman is weak to kryptonite" or "The king of France is bald." This will be the first problem raised by the move to meaning grounded in objective correlation: how do we explain the way we understand the meaning of linguistic entities that do not have any such correlate.
MER seems to arise pretty immediately in response to COR & OBJ. It also is related to an old problem, already raised by Plato (in the Theatetus?). Competing strategies (positing another extra-linguistic layer (Frege), (analyzing sentence to expose logical form that makes no such reference, i.e., one that relies on objectively correlative meaning (Russell) will arise in line with broader theoretical commitments (Frege's light of making sense of communication, Russell's in light of epistemic considerations).
(NTI) the problem of making sense of non-trivial identities e.g., "Hesperus is phosphorus" or "2+3=5". Why should these sentence be meaningfully different from the trivial identity claims "Phosphorus is phosphorus" and "5=5", if the entities on both sides of the equation are meaningful in virtue of the same objective correlates.
NTI was raised by Frege as a way of defending his logicist project, needing to clarify a universal sense of identity (follow up: Elucidations).
(UNT) ACCOUNT OF UNITY - Any theory of meaning will have to explain the holistic unity of the sentence. This can be exhibited by considering the difference between a list and a sentence: you can remove and add items from a list and it remains a list, however removing a term from sentence may cause it to cease to be a sentence. Since linguistic terms are as such ARB (there is nothing about linguistic entities as such that ensures unity), there must be some extra-linguistic ground for this unity.
Propositions "cannot consist of names alone; they cannot be classes of names"...For a class of names offers no prospect of supplying the resources to explain the unity of the proposition" NL 111.
Accounting for unity without supposing OBJ seems more easily accomplished: Locke, and Kant?,for instance, held that it was the mental act of judgment that 'holds together' the elements of sentence: unity as a mental, occurrent act. But since OBJ is not supposing linguistic meaning to depend on mental acts, it seems it is committed to the idea that their is an extra-linguistic objective unity in the world that grounds the unity of the sentence. This will also be related to Frege's context principle, I think. Wittgenstein takes it as a given that one cannot 'judge a nonsense', so if judgment constitutes the unity of the sentence, then nothing would rule out judging a nonsense.
If we observe that the (SPR) meaningfulness of a proposition cannot depend on its truth, otherwise we could not make sense of falsehood, then the problem of unity can be seen as the problem of accounting for unity as a tension between the requirement for objectivity (that the extra-linguistic unity must be objective) and that false sentences must equally be possessed of that same unity: so we require an account of unity that is somehow objective but also non-actual. Russell, who is most stringently attached to the actuality (to the point of indubitability) of any objective correlates will struggle most to make sense of this. Frege seems to depend on making sense of the unity of thought. While W will deny that a sentence has an extra-linguistic entity from which it gets its meaning.
F,R and W all see that they must account somehow for the form of the proposition, that it is not the heap of names or terms that can account for the fact that something is asserted, but essentially an orderly combining of them. (a cobb salad is not just all the ingredients of a cobb salad). Early, Russel inspired solution involved casting a proposition as an existential assertion about a complex. Frege used the notion of function but this does not help with understanding a proper analysis of the proposition (scrutiny of a proposition cannot reveal the function whose value for the argument it is.)
Here Wittgenstein must add the notion of form to the picture, which must be something radically different from a name.
(GDV) QUESTION GRAMMATICAL DIVERSITY - This is a question about whether, or how many, logical/grammatical types their are. Possibilities will be horizontal (names, predicates) and vertical (as in a hierarchy of types) and unclear (logical constants).
The main force of the argument for an undifferentiated horizontal grammar is made by Russell. He seems above all committed to the idea that any term must be able to be spoken of as subject and as quality. It is easy to see the complications involved of differentiating or defining the category of predicate without making them objects and thereby referring to something essentially difference than what we want to be talking about. Ordinary language seems to allow a single concept to change its form as needed - a throw, throwing, thrown, etc. Intuitively there should be a single concept that these make reference to, however different grammatical types will make it not only impossible for the surface connection to be made between a throw and throwing, but it will make it impossible to differentiate them at the grammatical level. Frege also considers this problem, where he needs to say something like, "...is a horse" refers to the concept horse.
Subscribe to:
Posts (Atom)