philosophy is for everyone
and not just philosophers

philosophers should know lots
of things besides philosophy

Philosophical Connections

Electronic Philosopher

Feature Articles

University of London BA

Philosophy Lovers Gallery

PhiloSophos Home

International Society for Philosophers

Five essays on the philosophy of language


To: Matthew A.
From: Geoffrey Klempner
Subject: Five essays on the philosophy of language
Date: 26th October 2010 14:48

Dear Mathew,

Thank you for your email of 28 September with your five essays for the Philosophy of Language program. Well done for completing this program. I will be sending your Pathways Certificate and summary report shortly. Well done also for producing work of a very high standard. This augurs well for your studies towards the Philosophy BA.

1. 'A gramophone record, the musical idea, the written notes, and the sound waves, all stand to one another in the same internal relation to one another of depicting that holds between language and the world' (Tractatus Logico-Philosophicus 4.04). Discuss.

You offer some good exposition here, and a clear account of what W. was trying to do in the Tractatus, and why he thought that the notion of 'picturing' was helpful as a way of explaining how propositions relate to the world.

When we say, as in your example, 'The fork is to the left of the knife' it seems intuitively clear that the relation between the words 'fork' and 'knife' maps on to their denotations in the world. The sentence is a fact, as W. states, and this is how it is able to represent a possible state of affairs, which is actual if and only if the two 'facts' match up or fit, one on top of the other.

Could *all* facts be like this? Let's leave ordinary objects behind (to which W's account applies 'metaphorically') and consider the underlying logical structure of combinations of names ('elementary propositions') representing combinations of simple objects, and the truth functions of elementary propositions which provide the ultimate analysis of our everyday language. The relational terms 'left' and 'right' belong to the spatial form of representation. In some possible universe, perhaps (e.g. a Cartesian universe where all physical properties are reducible to three-dimensional geometry) a complete description could be given in terms of such spatially ordered elements. But that seems a far cry from the physical world that we know.

Why does W. talk of an 'internal relation of depicting', and why is the example of the gramophone record etc. only an analogy?

The relation is internal rather than external because it is based on rules, or as W. describes it in the Tractatus, a 'law of projection'. In the case of music, the form of this law is not 'logical form' because a piece of music is not a proposition or made of propositions. Its form is unique, not shared by any other example of representation: musical form. Interestingly, something analogous to truth functions exists within musical form. The sign for crescendo, for example, does not specify exactly how much louder the piece should be played. Nevertheless, one could analyse it in a similar manner to truth functions, a series of instructions connected by 'or': any one of a range of possible values for the increase is acceptable, anything outside that range is unacceptable.

You say something puzzling: 'While it is training that allows the musician to interpret the picture of the song provided by the score, there seems to be no analogous training that explains how we understand the propositions of these natural languages.' The contrast you seem to be alluding to is between the fact that musical notation carries its structure 'on its face' whereas with natural language the structure is hidden. But is that really so? Isn't the problem the same, given that musical notation necessarily allows a degree of latitude in how the notes are to be played?

Ultimately, there is a difference: 'Since the entirety of our linguistic resources is used to picture the world, none of this linguistic resource is available to explain its relationship with the world.' This is why the musical example is only an analogy. It is language -- logical form -- which describes how musical notation 'works'; but there is nothing outside language to explain how language works.

2. Discuss the implications of the Private Language Argument

You have stuck to the question and considered what follows if we accept the validity of the PLA. However, we need to be clear what exactly this argument rules out. The term you use is 'epistemologically private experiences'.

I see a problem with this, however, in that it seems perfectly acceptable to say that there are things each of us knows about our inner states which, though communicable to others, are known by us in a unique way. I don't need to hear myself say, 'I am in pain', or observe my pain behaviour in order to know that I am in pain. W's response to this is to deny that it makes any sense to say, 'I know I am in pain', as this adds nothing to the statement 'I am in pain', or indeed to my observable pain behaviour.

The point, however, is that this is not *knowledge of an object*. What the PLA denies is that there is an object of my inner perception which presents an aspect only to me. The things we call examples of 'pain' in the actual world have dual or multiple aspects. You can *observe* someone else's pain.

Although philosophers use the term 'Cartesian mental events' as a synonym for 'private objects' I think that this is unfair to Descartes: in Meditation 1 he never considers the solipsist hypothesis. Even if all that exists is me and the evil demon, my mental experiences have two aspects, not one, the aspect they present to me and the aspect they present to the evil demon. In Berkeley's immaterialism, God becomes, in effect, a 'benign demon'. Idealism, in its various incarnations, is a tougher nut to crack than solipsism.

Russell and the earlier writings of the logical positivists (Carnap, Ayer) are a different case entirely. Here, Russell's 'sense datum propositions' or Carnap's 'protocol statements' seem to directly contravene the PLA. So why weren't Russell or Carnap solipsists? They ought to have been; but they succeeded in persuading themselves that solipsism can be ruled out on the basis of 'inference to the best explanation'. From the point of view of the PLA, this is mere wishful thinking.

3. Explain Russell's Theory of Descriptions, and discuss the claim that the sense of a proper name is equivalent to a description of the object which the name picks out.

This is a good go at a difficult topic. You explain the motivation for Russell's theory of descriptions, and argue successfully that if we take Russell's analysis to the extreme -- as indeed Russell himself did, allowing only names of sense data to be 'logically proper names' -- the result is solipsism, and Wittgenstein's PLA 'delivers the coup de grace'.

But let's backtrack a bit. The cluster of descriptions theory of proper names is still considered today to be a viable option. How do its proponents avoid the Russellian dead end? In his posthumously published book 'The Varieties of Reference' Gareth Evans argues for a Fregean account of our understanding of everyday singular terms, such as proper names of people or places, in terms of our practical ability to find our way around the world. The Fregean 'route to reference' or 'mode of presentation' is not necessarily reducible to a description or set of descriptions.

Descriptions enter the picture where we talk about objects beyond our acquaintance in this extended sense. Here, there is significant room for debate, between those who regard reference as capable of being preserved by a 'causal chain' of communication (as in Kripke's theory, and Evans original account which he recants in his book) and those who insist that knowledge -- albeit knowledge shared by the community of language users -- is necessary for successful reference.

I disagree with your analysis of 'Odysseus was set ashore while sound asleep', however.

You have, 'There is an x such that x is the Hero of Homer's Odyssey and x is son of Laertes and Antikleia and x was set ashore while sound asleep AND any y which is the hero etc. and the son etc. and set ashore etc. is equal to x.'

My analysis would be, 'There is an x such that x is the hero etc. and x is the son etc. and any y which is the hero etc. and the son etc. is equal to x, AND x was set ashore while sound asleep.'

What is the difference? In your analysis, if it turns out that Odysseus was not set ashore while sound asleep, then Odysseus does not exist, because you have defined 'Odysseus exists', in effect, as equivalent to the existence of a unique x which satisfies the three descriptions. Whereas on my analysis, if Odysseus was not set ashore while sound asleep it doesn't follow that Odysseus doesn't exist.

4. Assess the philosophical significance of the claim that to understand a proposition is to know its truth conditions.

You offer a useful exposition of Davidson in answer to this question, on which I won't comment. Although Davidson is known as the prime exponent of the view that meaning can be understood in terms of truth conditions, he was certainly not the first philosopher to put forward the claim that to know the meaning of a proposition is to know its truth conditions. Frege, despite his seeming commitment to a non-physical world of immaterial 'senses' sought to provide a truth functional analysis of natural language, and this commitment was taken up by Wittgenstein in the Tractatus.

It was Frege who first claimed that a proposition is a *function* whose value is 'the true' or 'the false'. Nor was Davidson the first philosopher to eschew 'meanings'. The analysis of language in the Tractatus is fully extensional. What Davidson achieved was the first feasible theory of natural language which relies purely on the powers of first-order predicate calculus. His analysis of 'saying that' was taken as the model for the philosophical inquiry into language by a generation of students.

What of the Davidsonian project today? The main battle lines appear to be drawn around the question of the adequacy of first-order predicate calculus, versus accounts such as Chomsky's which argue for the existence of a 'deep grammar' to natural language which cannot be reduced in this way. The question, in effect, is whether it is the task of the philosopher to investigate the semantics of natural language, or whether in the interests of seeking the 'best explanation' we have to admit the relevance of empirical inquiry into how language is actually learned.

5. 'Anti-realism about truth entails that 'reality' is our own invention. When it comes to deciding what to believe, anything goes, for according to the anti-realist there are no objectively right or wrong answers to our questions. It follows that there is no difference between reality and a mere dream.' -- Comment on this attack on the anti-realist theory of truth.

It is good to see that at the end of your essay you recognize the point I made above, concerning the difference between the solipsist theory, and Descartes' hypothesis of an 'evil demon'.

I don't agree with your defence of anti-realism, however. The possibility which is being envisaged here is that reality is in some sense reduced to *our* dream.

You do consider the challenge posed by concepts such as 'phlogiston' (another example would be 'witch') which seemingly 'have a meaning' in our public, inter-subjective world and yet fail to refer to anything in reality. You also take this further, in considering how our inter-subjective world would be viewed from the vantage point of the far future, where the concepts we regard as objective have no more validity than 'witch' or 'phlogiston'.

But what is the response to this challenge?

'According to the anti-realist there are no objectively right or wrong answers to our questions.' This isn't a criticism of anti-realism; it is what the anti-realist claims, in effect. To 'answer' a 'question' one pursues the appropriate method of investigation. Having made our 'decision' we recognize that we can always be 'wrong', but all this means is that further investigation might lead us to change our minds. There is no room, in the anti-realist account, for a notion of truth as correspondence, in the sense in which this implies the existence of truths which we can never discover or determine.

This isn't solipsism, but it is (or might appear to be) equivalent to a view which has consequences which are no less catastrophic: there is no 'world', there is only the story we tell about 'our world'. The anti-realist's response is, in the words of Richard Rorty, that the realist's 'objective world' is a 'world well lost'.

All the best,