On the Failures of Nonsense-Policing and Ordinary Language Philosophy

In this post I reflect on the failures of nonsense-policing and ordinary language philosophy, and the fact that notwithstanding these failures, paying critical attention to semantic issues is of central importance in philosophy, and in metaphysics as well as philosophy of language. 

The early Twentieth Century saw a surge of interest in the idea that the apparently intractable problems of philosophy come from misunderstanding language, and that philosophical questions are often unanswerable not because they are too difficult for us, but because they fail to make sense, or we fail to understand them properly.

This idea took various forms. The logical positivists attempted to give general criteria of meaningfulness using the notion of verification, the early Wittgenstein aims at a comprehensive, general theory of how language represent which leaves no place for genuine philosophical propositions, and more specific doctrines in logic and philosophy of language were used to dissolve particular problems.

Logical positivists attempted to formulate a criterion of meaningfulness in terms of the notion of verifiability. The basic idea is that a meaningful sentence must either be analytic (i.e. true in virtue of meaning) or verifiable in principle by experience. Philosophical sentences which are neither analytic nor verifiable experientially are therefore meaningless if this idea is correct, and works of metaphysics appear to be full of such sentences. As powerful as this idea seemed to the logical positivists, attempts to work it out in detail ran into trouble. There are also serious problems affecting even the basic idea; what about statements about minute happenings in the distant past, or events outside our light cone? (Such cases are more or less worrying depending on how one thinks about ‘verifiability in principle’.) And what about competing theoretical claims that do not appear to be analytic but where the contest does not appear to be settled by any possible experience? These arise in science as well as philosophy and we seem to decide about them on the basis of things like theoretical virtues rather than empirical tests. Thus the whole idea of experiential verifiability as a criterion of meaningfulness has been largely abandoned (at least in academic philosophy; it enjoys a vigorous half-life, along with Popper’s idea that truly scientific theories must be falsifiable, in the wider intellectual culture).

Wittgenstein in the Tractatus (a source of inspiration for the verificationists) attempts to say in general how meaningful propositions are constructed. In numbered, often cryptic remarks, we are given a beautiful, austere picture of how language represents. Elementary propositions are constructed out of names, and more complex propositions are constructed from them by simple logical operations. Anything which can’t be constructed in this way is not a genuine proposition. And many sentences in philosophical works fail to be genuine propositions. Wittgenstein’s proposal is that, with theses and questions in metaphysical philosophy, something akin to what is wrong with the sentence ‘Mortality is Socrates’ has gone wrong, only more subtly so.

(Wittgenstein’s later work is also replete with denials of sense, although it is up for debate what these come to and what they are based on. Assessing expressions for sense becomes a looser, more occasion-sensitive affair in the later work, as opposed to being based on some general theory of meaningfulness. Sense-denying aspects of Wittgenstein’s work, and interpreters of his work who emphasise them (e.g. P.M.S. Hacker), have come in for extensive criticism, and are seldom echoed in more recent contemporary philosophy. I see them as an unfortunate holdover from the early work, and as marking a place where Wittgenstein’s ideas require both revision and further development.)

Apart from general criteria of meaningfulness and overarching theories of what can be said, logical doctrines about particular parts of language have also been used to dispense with philosophical problems by denying sense. A famous example is the doctrine, defended by Bertrand Russell and wielded by many later writers (a prominent example being Rudolph Carnap), that ‘exists’ cannot be used with proper names. This is a kind of modern logical version of Kant’s doctrine that existence is not a predicate. From a contemporary point of view, this seems hopelessly revisionary. What we ordinarily call ‘proper names’ can meaningfully be put together with ‘exists’, and logically minded philosophers must accommodate this fact as best they can. 

The above views have in common that they are supposed to be based on the technical foundation of modern logic. Another influential Twentieth Century development of the idea that a better understanding of linguistic meaning is methodologically crucial is the Ordinary Language Philosophy (OLP) movement, as spearheaded by J.L. Austin. In a way it involves a kind of nonsense policing, but there is no general, technical policy that is being enforced; the OLP nonsense police, if that’s what they are, know that they’re operating in a complicated society and will require good judgement. This no doubt enhanced OLP’s plausibility in the wake of logical positivism’s failure.

I once heard from a mentor that in the 1960’s there was a feeling in the air that the traditional problems of philosophy would soon be cleared up by means of OLP. That this hasn’t happened should be uncontroversial. Why it hasn’t happened is more controversial, but we can speculate. One factor seems to be that the urge to use language in extraordinary ways is deep-rooted. No amount of attention to ordinary usage can stop it. Metaphysicians have become more self-conscious about the fact that their use of key terminology may depart from ordinary usage, and have found ways to persist in it nonetheless. (For example, in response to arguments by Eli Hirsch to the effect that non-commonsensical claims in ontology simply couldn’t be true, since they fly in the face of metasemantic wisdom - charity demands that we interpret natural language sentences so they aren’t radically false - Theodore Sider has argued that non-commonsensical metaphysical views should be regarded as being stated in Ontologese, a special non-ordinary language. For the Hirsch-Sider debate, see (in dialectical order) Hirsch (2010) (a collection of essays), Sider (2009), Hirsch (2008) and Sider (2014).)

Another factor may be a lack of technical sophistication: philosophy can get quite technical, but OLP tends not to come along for the ride. (Perhaps proponents of OLP could have done better here, either by coming along for the ride in some fashion or by forestalling misguided research programmes.) OLP at its most interesting can also become quite idiosyncratic and fragile, in contrast to a more recent trend of international and interpersonal cooperation. One of the most intriguing developments of OLP, the work of Stanley Cavell, typifies this tendency: you see all this sophistication, all these interesting scruples, but you have to be a special kind of person in a special kind of culture to appreciate it. Interesting as it may be, such work can be difficult to make use of in the sort of thing that Bertrand Russell called ‘technical philosophy’. Above all, the ordinary language philosophy of the mid Twentieth Century has simply been sidelined by more vital developments in connection with which it appears at least to have little to offer. Quine, Putnam, Kripke, Lewis and many others come on the scene, we get a whole raft of topics and problems to work on, and ordinary language philosophy falls by the wayside. 

The failures of nonsense-policing and ordinary language philosophy have had the unfortunate consequence of leading to complacency about sense in contemporary philosophy; a kind of presumption of innocence, where once basic standards of clarity have been met, we tend to take it for granted that philosophers’ utterances are meaningful and that we understand well enough what they mean. According to this way of thinking, the important work in philosophy tends to consist, not in inquiring not into the sense of expressions, but rather into which claims are true.

But the failure of nonsense-policing should not be taken to suggest that the analysis of meaning is not of central importance in philosophy. There may not be much of a future for denying meaning to animating words and phrases, but there remains a need to be critical and inquisitive about what they do mean. We are often being critical in this way even if we aren’t explicitly talking about meaning, or aren’t doing so systematically. But to get further we need to be able to be self-conscious and systematic, without losing crucial data that doesn’t fit well with the kind of thin, reference-based conception of meaning that now seems to hold sway again in philosophy.

References

Hirsch, Eli (2010). Quantifier Variance and Realism: Essays in Metaontology. Oxford University Press.

Hirsch, Eli (2008). Language, ontology, and structure. Noûs 42 (3):509-528.

Sider, Theodore (2009). Ontological realism. In David John Chalmers, David Manley & Ryan Wasserman (eds.), Metametaphysics: New Essays on the Foundations of Ontology. Oxford University Press. pp. 384-423.

Sider, Theodore (2014). Hirsch's Attack on Ontologese. Noûs 48 (3):565-572.

On the Need to Distinguish Knowledge of Essential Properties from the Knowledge That They Are Essential

In one of my favourite papers about modal metaphysics, Rosen's 'The Limits of Contingency', Rosen develops the idea that there are two sets of modal notions which are apt to be conflated under the heading 'metaphysical modality'. Roughly: a correct conceivability modality, and a more properly metaphysical modality on which some things which are correctly conceivable are in any case not really possible.

In a recent paper, 'Correct Conceivability and its Role in the Epistemology of Modality' whichin a new book Les Principes Métaphysiques published by the College de France, Robert Michels considers an objection to Rosen's idea of a correct conceivability modality construed as giving an account of how considerrations about conceivability can takes us from non-modal to modal knowledge. I think this objection is based on a misunderstanding, as I will explain in this post. 

Let me quote from Michels, first to set the scene, and then to see the objection and the main responses Michels considers.

Correct Conceivability

The basic idea

10The correct conceivability-approach presupposes the Kripkean standard view of metaphysical modality and accordingly has to rule out modal errors, cases in which a state of affairs is conceivable, but not metaphysically possible. How this is done is nicely explained in the following mock-quote of Rosen’s “others”:

If the ancients could conceive a world in which water is an element, this is only because they were ignorant of certain facts about the natures of things. In particular, it is because they did not know what it is to be water. They did not know that to be water just is to be a certain compound of hydrogen and oxygen – that to be a sample of water just is to be a quantity of matter predominantly composed of molecules of H2O. This is not to say that they did not understand their word for water. But it’s one thing to understand a word, another to know the nature of its referent. The ancients could see no contradiction in the supposition that water is an element because they did not know that water is a compound by its very nature. But we know this; and given that we do, we can see that to suppose a world in which water is an element is to suppose a world in which a substance that is by nature a compound is not a compound. And that’s absurd. (Rosen 2006, p. 22-23)

11The idea is hence that in order to avoid modal error, conceivability needs to be supplemented by knowledge of the natures, or equivalently, essences of relevant entities, in this case the essence of (the property of being) water. Equipped with this knowledge, the conceiver is able to detect that the assumption that water is an element together with essential truths about the relevant entity entails an absurdity.

12The correct conceivability-approach hence gives us a simple and elegant explanation of why we are apt to make modal errors and a recipe for ruling them out. We tend to commit modal errors because we can conceive of states of affairs which are ruled out by relevant essences. To avoid these errors, we have to let our ability to conceive be guided by knowledge of the essences of relevant entities.

An objection to essence-based conceivability approaches and how it can be addressed

13There is a rather obvious objection to the correct conceivability-approach which should be mentioned: The approach crucially relies the conceiver having knowledge of essence, but essence itself is a modal notion. Doesn’t this mean that the notion of correct conceivability cannot answer the core question of modal epistemology, the question of how we can acquire modal knowledge?

• 5 See for example Lowe (2012), Hale (2013, ch. 11), Tahko (2016), Tahko (2017). For a recent critical (...)

14

(End of quote.) 

Michels considers a number of responses, but not the following simple one which I think is correct.

Yes, the approach relies on the conceiver knowing that some thing has a property, where arguably that property is in fact an essential property of the thing. But the approach does not rely on the conceiver knowing that the thing essentially has that property.

The idea is not that people learned that water is essentially H2O and then on that basis, saw that water being simply an element is not correctly conceivable. Rather, they just learned that water is H2O, and then on the basis of their grasp of how the concept of water works, saw that water being an element is not correctly conceivable. The knowledge of essence, just like the knowledge of necessity (Fine taught us to disinguish the two), may perhaps also be accounted for along similar lines. But that's by the by. This objection simply doesn't get off the ground as far as I can see. 

This sort of confusion is not unique to Michels's paper. Phrases like 'knowledge of essence' and 'knowledge of necessity' are apt to blur this important sort of distinction, leading to mistakes in the epistemology of modality. 

References

Michels, Robert. Correct Conceivability and its Role in the Epistemology of Modality In : Les principes métaphysiques [en ligne]. Paris : Collège de France, 2020 (généré le 18 septembre 2020). Disponible sur Internet : <http://books.openedition.org/cdf/8079>. ISBN : 9782722605350. DOI : https://doi.org/10.4000/books.cdf.8079.

Rosen, Gideon (2006). The limits of contingency. In Fraser MacBride (ed.), Identity and Modality. Oxford University Press. pp. 13--39.

On Family Resemblance Concepts

I want to clarify some aspects of the celebrated Wittgensteinian idea of a family resemblance concept.

Are all family resemblance concepts such that no single feature is shared by all the things to which the concept applies?

Some of Wittgenstein's formulations, and explanations based upon them, would suggest an affirmative answer. But I think that a more general, useful notion of a family resemblance concept should not require this.

It should disqualify a concept for family resemblance status that some feature happens to be common to all the things that fall under the concept. It may be that this feature does not sufficiently characterise the things as falling under the concept. That is, just because you can give necessary conditions for the application of a concept, does not mean its application is not based on the kind of criss-crossing network of similarities that Wittgenstein has us imagine. This passage gets it right:

we extend our concept of number as in spinning a thread we twist fibre on fibre. And the strength of the thread does not reside in the fact that some one fibre runs through its whole length, but in the overlapping of many fibres. (PI §67)

This makes it clear that there may still be a single fibre running through the thread. It's just that that alone isn't responsible for the strength of the thread.

Are all concepts which do not admit of non-trivial analysis in terms of necessary and sufficient conditions family resemblance concepts?

It seems to be a necessary condition on family resemblance concepts that they do not admit of non-trivial analysis in terms of necessary and sufficient conditions. But it doesn't seem to be a sufficient condition. That is, it seems that there are concepts which do not admit of this sort of analysis and yet are not what we would think of as family resemblance concepts. For example:

• Certain primitive, i.e. undefined, concepts in theoretical use are not family resemblance concepts. For example, the concept of a lever (insofar as it isn't defined).
• A concept that is undefined but very closely tied to experience, e.g. the concept of red.
• Concepts which by design only apply to one thing, such as the concept of God - or, for that matter, just your concept of some particular person you know. (The whole idea of 'individual concepts' is a bit neglected though, and may not strike the reader as being in good standing.)

These do not seem to count as family resemblance concepts, because while they do not admit of non-trivial airtight definitions, there does not seem to be the sort of heterogeneity among the things that they apply to that characterises family resemblance concepts as discussed by Wittgenstein.

This last category, however, invites another question...

Diachronic family resemblance concepts?

It would seem that the concept of a particular person can't be a family resemblance concept in the sense of many different things falling under it not because of a common feature but because of overlapping similarities, because at most one thing does fall under it. But if we consider the individual through time, we start to see the possibility for something like the family resemblance idea applying to the concept of a particular person.

Is 'family resemblance concept' a family resemblance concept?

I think there's potential for a negative answer. At least, it seems to me we can give a precisified version of the idea that is not itself a family resemblance concept. On the other hand, perhaps such a thing would fail to capture the idea in full.

Opportunism and Fastidiousness in Language

Many, probably most, intelligent speakers and thinkers are linguistic opportunists. They treat words and phrases as tools to convey things that they are already thinking in proto-linguistic or rough linguistic form, and if some construction seems likely to do the conveying job, they use it. 

Linguistic opportunism is higher in private communication than in public, and higher in informal writing than in formal writing. The opposite of linguistic opportunism is linguistic fastidiousness. Analytic philosophy is characterised by high linguistic fastidiousness, while other types of philosophy are often linguistically opportunistic.

Certain moves that may at first seem like moves of fastidiousness belie an underlying opportunism. For example, an linguistic opportunist may abandon a formulation and self-correct to a formulation which lets their underlying message come through clearly, where a stickler may take the time and effort to clarify the sense in which what they said was correct although potentially misleading.

(I heard an example of this in an episode of Dan Carlin's Hardcore History podcast, but I can't remember the details. Carlin is an interesting example of a fairly fastidious opportunist. As I see it, his underlying goal makes him fundamentally an opportunist. His goal is to paint pictures with words, to move the listener, and to convey history. But he often becomes self-conscious about language. Nevertheless, this is all kept in the service of the goal, and so does not go into niceties that the true stickler would go into.)

What drives the stickler? In large part and in many cases, their motivation takes the form of an intrinsic interest in language and fine distinctions. (But remember, one may ask why such an interest should ever arise in the first place.) As someone with marked stickler tendencies myself - tendencies which have at certain points in my development reached nearly pathological levels - there are many times when I have become interested in some linguistic or conceptual issue raised by some construction that has been used, where this definitely only detracts from the task of staying on-message. But sticklerhood is not always like this. Sometimes it is motivated by real, felt concern with the ongoing adequacy of the linguistic resources to the task at hand. 

Even the purer form of sticklerhood, which unambiguously cuts against the grain of achieving the purpose of the discourse from which it arises, may have a justification beyond the intrinsic interest of reflecting on language and concepts. The stickler may be seen as concerning themselves with utility too. Not with having the right tools for the task at hand, but with maintaining a large, powerful toolkit for whatever may come next. And sometimes this goal may be in conflict with an opportunistic strategy which would work well for current purposes.

It may seem self-evident that the linguistic opportunist is a friend of linguistic innovation, the fastidious stickler a foe. But this is too quick. By maintaining order and consistency, the stickler helps to cultivate an environment in which buildings can be erected which, if the opportunist held sway, would crumble into confusion. 

Forthcoming in Thought: The Accident of Logical Constants

The Accident of Logical Constants

Work on the nature and scope of formal logic has focused unduly on the distinction between logical and extra-logical vocabulary; which argument forms a logical theory countenances depends not only on its stock of logical terms, but also on its range of grammatical categories and modes of composition. Furthermore, there is a sense in which logical terms are unnecessary. Alexandra Zinke has recently pointed out that propositional logic can be done without logical terms. By defining a logical-term-free language with the full expressive power of first-order logic with identity, I show that this is true of logic more generally. Furthermore, having, in a logical theory, non-trivial valid forms that do not involve logical terms is not merely a technical possibility. As the case of adverbs shows, issues about the range of argument forms logic should countenance can quite naturally arise in such a way that they do not turn on whether we countenance certain terms as logical.

Pre-print available at PhilPapers