The Meaning of Meaning

What do we mean when we ask, "What do we mean?" This is a problem within philosophy of language. Philosophy of language in my view reigns supreme over all other branches of philosophy because it undergirds all fields of knowledge: all scholarly disciplines, with the possible exception of mathematics, rely on language to explicate knowledge and theories and to communicate knowledge and theories from one person to another. To understand the foundations of all knowledge we need to enquire into the nature of language. The purpose of this essay is to present a theory within philosophy of language, a theory of meaning. It picks up from the previous essay I published in this blog some time ago. Because in that essay I tried to cover too many different topics in the same post, not all of which were closely related to the core topic, it may have seemed to meander. In this post I am going to stick more closely to my main topic: the meaning of 'meaning'. This essay will be longer than that typically written for a university course but will, of course, still not do justice to the topic. My main sources for writing on this topic are the cluster theory of proper names first proposed by John Searle, Kripke's seminal lecture series "Naming and Necessity", and the article "Word Meaning" in the online Stanford Encyclopaedia of Philosophy, although I shall reference other works. I actually have a novel contribution to the conversation to make, an idea I presented in the previous essay but which I think I can defend more robustly here. I shall finish the essay by considering the notion of 'truth'.

Often when presenting a philosophical argument, we must begin by defining the central terms or stipulating how they will be used. The salient word here is 'meaning'. English language speakers use the words 'mean' and 'meaning' routinely. We say, "What do you mean?" or "I didn't mean to do that," or "I didn't understand what the poem meant" or "What does that word mean?" or "What is the meaning of life?" Nevertheless ordinary speakers may have trouble defining the word 'meaning'. The online Miriam Webster dictionary defines 'mean' as follows: "to have in the mind as a purpose : INTEND [...] or to introduce a phrase restating the point of a preceding phrase [...], to design for or destine to a specified purpose or future; to serve or intend to convey, show, or indicate : SIGNIFY [... ]; to have importance to the degree of [...] to direct to a particular individual." Although the dictionary definition does not clearly signpost this, meaning has always seemed to me and hopefully to many of my readers a mental phenomena. The meaning of the word 'cat' is, to speak a little loosely here, some kind of mental representation or concept competent English speakers associate with the phonological and graphological sign 'cat'. Although we also associate the term 'meaning' with 'intention' I shall not be using the term 'meaning' in this way here but rather as denoting some kind of mental correlate of a sign of any sort we perceive in the world.

This way of parsing the meaning of 'meaning', as a mental correlate of some sign in the world, is, I think, a twentieth century development. If we go all the way back to Plato, we find the doctrine that the word 'horse', for instance, was supposed to be associated with the Idea of Horseness. Ideas, for Plato, were, to quote Wikipedia, "non-physical, timeless, absolute, and unchangeable essences of all things, of which objects and matter in the physical world are merely imitations". The realm of Ideas is eternal and mystical; it may exist in the minds of people but also somehow elsewhere somewhere. I suspect that Plato didn't have a solid conception of the human mind and consequently didn't consider the possibility that different people might possess different ideas, or that ideas can change over time both at the personal and community level. Although we still use the term 'Ideas' for Platonic Forms today, this term may to many of us seem misapplied because so many of us now consider ideas to be mental phenomena. For Plato 'horse' is both a word and a kind of thing to be found in the world, a set of things that are imitations of some prototypical ideal Horse in the realm of Ideas. Thus from the beginning there was confusion between the meanings of words and the types of things that are found in the world.

Of course philosophy is complex and different philosophers have understood the word 'meaning' and its analogues in other languages differently. For instance the very famous philosophers John Locke and David Hume seem to have viewed the meanings of words as being mental pictures and so did indeed interpret meanings as mental phenomena. However historically and today meaning has often been tied to reference. By the nineteenth century, philologists had realised that the meanings of words can change over time but philosophers didn't recognise the philosophical import of this. Within philosophy a distinction came to be drawn between two different types of word: proper names and common words, the latter being generic nouns, verbs, adjectives, and adverbs. Both types of word have meanings, and in fact in this essay I will not draw a sharp line between the two. But historically the two types of word have been treated quite differently.  I shall focus first on proper names. In 1843, John Stuart Mill argued that the meaning of a proper name simply is its referent, that there is no 'semantic' content to a name like Dartmouth or Aristotle. It was not until Frege, in 1892, proposed the distinction between 'Sinn" and "Bedeutung", usually translated as "sense" and "reference", that we begin to get a distinction in philosophy between the meaning of a proper name and its referent, when we start to tease the two Platonic conceptions of ideas apart, and even then this new philosophical paradigm is still embryonic.

The next major development in this line of philosophical thought occurred in 1905 when Bertrand Russell proposed that the semantic content of a proper name is a definite description. For instance, the semantic content of the name "Aristotle" is (partly) "the teacher of Alexander the Great". The notion here is that by virtue of this definite description we can pick out uniquely the object the word refers to. Of course there is more than one definite description that can be associated with "Aristotle" and I don't think Russell explained why we should pick any one definite description rather than any other to be the semantic content of a name. There are other problems with their theory which I will not spell out here, although if the reader is interested they can perhaps watch the video "Proper Names by John Searle" on Youtube. This brings me neatly to John Searle. Searle criticised both the tradition associated with Mill and the one associated with Frege and presented an alternative theory. According to Searle, a proper name is associated with a cluster of descriptions, none of which are necessary to the meaning of the word. In fact Searle thought that the looseness associated with the meaning of proper names, the fact that no single descriptor is necessary, is what makes them so useful: proper names are tags to which people can attach descriptions that can change over time and, although Searle does not say this explicitly, his theory coheres with a view of the world in which different people can understand the meanings of words differently and of a world in which the meanings of words can shift.

Searle does not use this type of argumentation in his famous essay but I believe I can begin to elucidate my own theory, a theory that follows directly from Searle's cluster theory, by invoking the notion of what we might call a conversational interaction. Suppose Jack and Jill are discussing philosophy and Aristotle comes up. Suppose Jill knows only a little about Aristotle: in her mind the cluster of descriptions might be only "famous Greek philosopher", "student of Plato", and perhaps some mental pictures of a bearded bust. Jack knows a little more: he knows for instance that Aristotle came up the Lazy Man argument. He tells Jill that Aristotle was one of the first known philosophers to propose an argument against predestination, that this argument is called the Lazy Man argument, and then goes on to summarise it. Jill has learnt something new, and if she believes Jack and remembers what he has told her, she has added another descriptor to the term "Aristotle", specifically "deviser of the Lazy Man argument". Of course, Jack, in telling Jill about the Lazy Man argument, is presupposing that Jill has at least a little familiarity with "Aristotle". However if Jack and Jill associated the exact same set of descriptors with the term "Aristotle", it would be impossible for Jack to tell Jill anything new about him. We could also imagine a situation in which Jill associates an incorrect descriptor with the term "Aristotle", perhaps that he died from drinking hemlock, and that Jack rectifies her understanding by informing her that she is thinking of Socrates. Through conversational interactions, people are constantly relaying to each other information that is new to the person hearing or reading it, and this seems to be one of the primary purposes of language. If everyone possessed the exact same understanding of language and of the world as everyone else, we would never need to talk to each other, never be able to tell anyone anything new. As another example, an example inspired by Frege, many casual astronomers throughout history may not have realised that the Evening Star is identical to the Morning Star until someone told them.

Searle's theory seems to me very close to the correct theory but, unfortunately, in 1970, Saul Kripke delivered a series of lectures known as "Naming and Necessity" and completely derailed philosophy. Kripke has been enormously influential but I believe his influence has been terribly damaging. In these lectures, Kripke repeatedly insists that when we are using a proper name like Aristotle, there is no semantic content but that we are simply referring to that man. He offers no argument for this contention but simply assumes his audience shares his intuition. Kripke attempts to debunk the cluster theory of proper names; in doing so I believe he misrepresents it. In his view, the cluster theory seems to imply that everyone has got together and, through something like a vote, picked out a set of descriptors that should be associated with a name like "Aristotle" and that these descriptors are then to be treated as both necessary and a priori. I believe that Kripke misunderstood Searle, perhaps because Searle didn't go as far as I intend to in this essay. Obviously Searle is not suggesting that some kind of vote is taken before anyone can start using a word like "Aristotle" or that any of the descriptors associated with "Aristotle" are necessary or known a priori. Jack assumes that Jill has some understanding of the name, an understanding he assumes most people have, but he does not assume that Jill was born knowing that a descriptor associated with "Aristotle" is "famous Greek philosopher". He assumes, as we all do when talking about Aristotle, that the person we are talking to has learned this sometime during his or her life. So it cannot be known a priori. Nor is any descriptor necessary: Jack can meaningfully say to Jill "Aristotle wasn't Greek" if he, perhaps, means to suggest that he believes Aristotle actually immigrated to Athens from somewhere else around the Mediterranean. If it is necessary that Aristotle be Greek, Jack's statement would not be meaningful to Jill but, I contend, Jill will still find it meaningful because there is still considerable overlap between the sets of descriptors both associate with the name "Aristotle". Thus none of the descriptors people associate with proper names are necessary either. Rather, when talking to someone else, one makes assumptions about what the other might or might not know; one assumes enough overlap in one's own and the other's understanding of words that we can meaningfully communicate but we do not assume the overlap is complete. If the overlap were complete, Jack would have nothing new to tell Jill about Aristotle – but if there was no overlap at all, the two would not even share the same language and would be unable to communicate at all.

So far I have been discussing proper names but now we can bring in another type of word: common nouns. (The discussion could be extended to verbs, adjectives, and adverbs as well but I shall not do so in this essay.) When talking about the meanings of common nouns, it seems we need a distinction between the essential and the accidental, and a closely related distinction, the necessary and the contingent. We might say that it is essential, necessary, to the word "horse" that it denote something that has four legs but that the colouring of any particular horse is contingent, accidental. The belief that common nouns have both necessary and sufficient conditions goes back to the ancient Greeks again, is of a piece with Platonic philosophy, and is still central to much modern philosophy. The difference between a proper name and a common noun is that with a proper name, like Aristotle, not only does the name only refer to a single object, it seems also that we could theoretically itemise an indefinitely long list of necessary properties associated with it whereas with a common noun the list is truncated. Some properties are held by all members of the class, are necessary to all of them, while some properties might be held by one member of the class and not by another. The distinction between the essential and the accidental is so entrenched that those of us who have been introduced to this idea find it difficult to imagine disposing of it.

A long time ago, however, I found an essay on the Internet, a essay I believe to be a famous essay by a famous philosopher, challenging this distinction. (I have been unable to relocate this essay so cannot name its author here.) In this essay, the author argued that in fact when we say some characteristic of a thing is essential to it, we are really just saying that it is a characteristic we find it almost impossible to imagine the thing not having, something we are very unwilling it give up. Consider: today we say that it is essential to the word "water" that it denote "a collection of H20 molecules" but the fact that water is constituted by H20 molecules was not discovered until 1811 and people had been happily using the word "water" and its synonyms in other languages for thousands of years meaningfully before this without knowing it. There may be many people in the world today who happily use the word "water" meaningfully without knowing that it is 'essentially' H20. One may be able to mount a persuasive argument that water ought to be regarded as H20 essentially but, if people can use the word "water" meaningfully without knowing this, it suggests that knowledge of this 'essential' property is not required for people to use this word meaningfully. We might also consider that the word "bachelor" is supposed to essentially denote "an unmarried man" but it is possible to imagine some hermits who have lived in the wilderness all their lives, who have heard the word "bachelor" but do not know its dictionary definition and believe it to mean something quite different. It seems, surely, that they can use the word 'bachelor' meaningfully when talking with each other even though they do not know the dictionary definition – although they might then meet people who tell them the dictionary definition and bully them into accepting it. Not only can the essential properties associated with common nouns vary among people, they can vary over time. Of course some properties seem, in the world we live in, to go together: chordates seem always to also be renates. However an unknotting of this particular puzzle, the puzzle of why natural kinds seem to exist in the world, why properties seem to come together in bundles, does not belong to philosophy of language bur rather to ontology. The aim of this essay is to discuss the nature of meaning and its mental character.

Like proper names, common nouns have historically been viewed as inevitably involving reference. There is however, to my mind, a knockdown refutation of the idea that meaning necessarily involves reference. Picture Jack and Jill again. Suppose Jill tells Jack, "This morning I saw a unicorn in my back garden." Now Jack, knowing that unicorns are mythical, may wonder if Jill hallucinated, but even so he would still find her statement meaningful. Even though the word "unicorn" has no referents, it would still likely have a meaning for Jack. Jill is telling Jack that she saw a pure white horse-like creature with an ivory horn protruding from its forehead in her garden. Contrast this little story with another. Suppose Jill says to Jack, "This morning I saw a thiggle in my back garden". Of course, no one knows the meaning of this word "thiggle" because I have just made it up. The word will not be entirely meaningless to Jack however because he will assume that whatever a thiggle is, it is something that could turn up in a back garden and the appearance of which is surprising enough for Jill to mention it. It will have little more significance to him than this though and so Jack could justifiably plead ignorance and say to Jill, "What's a thiggle?" To which Jill might say, "It's what my mother called a fantail." Or, "It's what my sister called a pixie." Either way Jill will have furnished Jack with additional descriptors, additional meanings, for the word "thiggle" My general thrust here is that, although a word like "thiggle" has very little meaning to competent English speakers, words like "unicorn", "elf", "dragon", "vulcan" and "vampire" do have meanings even though they all lack referents. Searle makes a similar argument in the context of proper names. He points out that from Millian perspective, the word "Zeus does not exist" must be parsed as "The name 'Zeus' has no referent" but that throughout history and even today there are people and animals called Zeus and so the proposition "The term 'Zeus' has no referent" is false even though most people would regard the proposition "Zeus does not exist" as true. This supports the claim that the word "Zeus" should be considered a tag for a number of descriptors that most people associate with the name "Zeus", such as "chief of the Greek gods", "father of Hercules", "husband of Hera", and "wielder of thunderbolts". The words we ascribe to mythical beings such as those listed above, unlike the word "thiggle", must surely be regarded as meaningful even though they lack referents. They figure in the mythical and literary treasure house of English speakers' linguistic history.

I arrive now at the crux of the essay. What do I mean by 'meaning'? The view I shall propose here is more indebted to linguistics and cognitive science than mainstream philosophy, and is supported by the success of AI developers in creating large language models like ChatGPT based on huge neural networks. ChatGPT I think passes the Turing test. I claim that the human mind can be considered an enormous network of interconnected nodes. Every word in a person's lexicon is connected to other words. For instance, the word "cat" is contrasted with "dog" and also contrasted with the word "cab". The word "bachelor" is connected to the phrase "unmarried man". This suggests a Structuralist perspective, a la Saussure, but unlike Structuralists, I claim that the nodes are not all verbal, but may also be sensory: the word "cat" in most people's minds also often conjures up images of cats performing catty activities, is connected to such memories or imaginings. An enormous proportion of a person's knowledge is non-verbal. We can recognise the difference between Emma Stone and Melinda Clarke even if we have trouble verbally articulating their defining features; we can distinguish between a London and New York accent even if we know nothing about non-rhoticity and rhoticity; we can pick which pop song is playing on the radio from the first couple of bars even if we know nothing about music theory and are not pitch perfect. The fact that a great deal of the knowledge a human possesses is non-verbal is, when you think about it, obvious. If I tell you, "Close the door!" and the meaning of the word "door" were entirely verbal, you would not be able to accede to my request. You need some kind of visual idea of what a door is to know what I mean; visual imaginings of doors form part of the meaning of the word "door". The relation between nodes may be extraordinarily complex. It enables humans for instance to make statement like "Mt Everest is bigger than a shoebox", a sentence that in no simple way involves either purely verbal or visual nodes, seems to somehow involve conceptual nodes or some kind of encyclopaedic knowledge; I have heard statements like this can trip up ChatGPT.

From this perspective, it seems likely that in the human mind there is, at a fundamental level, no clear distinction between essential and accidental properties, but rather a reticulated dispersal of meaning from one node to those proximate and then those more distant. Eleanor Rosch has argued that when we think of a "cat", we imagine a prototypical cat and then tend to think of things as either more or less cat-like, rather than drawing a sharp line between "cats" and "non-cats" based on some criteria, as would seem to be the case if the meanings of words were constituted by essential properties. Rosch's proposal seems plausible to me.

The second vital point to note about language is that it is learnt. Of course this is obvious. Chomsky argued that the fundamental syntax of all languages is innate, perhaps genetic, and this is why he invented the idea of Universal Grammar. But no one today can seriously take the Platonic position that the semantics of a language is innate. No baby is born knowing the meaning of the word "unicorn". It takes years for a child to learn the meaning of the words used by those around her and then to start speaking herself. Even more importantly, the language the child learns is that spoken by those raising her. The obvious fact that different linguistic communities possess different languages conclusively proves the claim that the semantics of a language is learnt. Infants are sponges, somehow picking up on how others around them use language and conforming to the lexical rules of those around them. The ability of small children to learn so quickly their mother tongue is, when you think about it, astonishing. By the time we are adults, we tend to understand the meanings of all common words in almost exactly the same way as those around us do. But even as adults we can continue to learn and I claim that even as adults we are often appending new descriptors, new meanings, to the words we know and even learning new words. For example, I learnt this week about the new film Furiosa and can attach to this proper name, which I did not know before, descriptors such as "latest film in the Mad Max series", "film featuring Chris Hemsworth in a starring role", and "set like the earlier instalments in Australia". It is not just lexical knowledge that is continually developing but all knowledge. Suppose Jill takes Jack to a new cafe in East Auckland, one he has not visited before. As a result Jack may well commit to memory the geographic location of this cafe, a fact that is not linguistic but spatial or topographical. Supposing that knowledge is somehow actualised in the brain, the lessons of neuroscience become pertinent: it is well established that a person's brain evolves enormously from infancy through childhood and adolescence and that the frontal cortex is not fully developed until a human is about twenty-five. But even those of us, like me, who are nearing middle age can continue to learn new information, albeit with much more difficulty than we did as children, and, if materialism is true, this suggests that the brains of adults are still constantly changing (although to a lesser degree than they did when very young). This is not to say, of course, that a person is capable of remembering everything he or she has learnt or been told over the course of his life; it is probable that although we might gain new descriptors for terms we also lose others. The capacity of the human mind is not infinite. 

The picture I am drawing is that 'meaning' is related to knowledge and that knowledge is an enormous network of interrelated signs, stored images and sounds, concepts, and so forth in a particular human mind. The meanings of words are not isolable units. You might know that "Aristotle" is "a Greek philosopher" but even then you cannot really know the meaning of the name "Aristotle" without also having at least some vague understanding of the terms "Greek" and "philosopher". It might be that even if you do have some understanding of the term "philosopher" you may not be able to distinctly articulate the meaning of this word – but you might connect it strongly in your mind with people like Rene Descartes and Bertrand Russell, less so with Darwin and Einstein, and not at all with someone like Donald Trump. You might say of a friend, "Amy is a bit of philosopher," if she occasionally says something deep. (To reiterate, not all knowledge is verbal or capable of verbal expression.) To try to delimit the boundaries of a word's meaning by invoking necessary and sufficient conditions ends up a kind of question begging. It seems moreover that the meanings a person attaches to a word are implicit, can only be inferred from the way the person speaks and behaves, that they cannot be accessed through introspection.

The theory I am advocating for here is related to meaning holism and I encourage the reader to look up the Stanford Encyclopaedia entry on this theory. However my own theory departs from the holism defended by Quine among others. Traditional holistic theories of meaning often seem to suppose that the network of meanings in the mind of one person must be either entirely identical with or entirely different from the person she is speaking to; this is wrong. As I argued above, it seems that when one talks to someone else, one assumes a great deal of overlap but not complete overlap. If the networks are identical, there would be no point communicating at all and if the networks are entirely different, communication would impossible. We constantly make assumptions about what others know when speaking to them. A second way my own theory departs from traditional holistic theories is that these theories suppose that if the meaning of one term changes it alters the meaning of every single other term in the network. But we can quite simply suppose that with the addition of a new descriptor, or adoption of a new belief, a change propagates to related nodes in the network without also supposing that every single term in the network acquires a new meaning. Dropping a stone in the Hauraki Gulf will probably not produce a ripple that will travel all the way to South America. If Jill learns that Aristotle came up with the Lazy Man argument it may affect her understanding of terms like "Aristotle", "philosophy" and, say, cause her to alter her beliefs concerning how long arguments about the existence of free will have been raging – but it won't affect her beliefs about the location of the Britomart train station. Finally, traditional holism has trouble explaining how one learns a language to begin with. Consider – a child's first word may be "No". If some holistic theory of meaning is true, how can the child know the meaning of this word "No" if presumably she knows no other word? However we can go some way towards solving this puzzle by recognising that the child has been observing the way others around her use this word, their body language, and the types of situations in which it is used. Quine and other proponents of meaning holism assume that the network consists entirely of words but, as I argued above, the network comprises in addition an enormous amount of perceptual experience and perhaps some elementary concepts. A network is in place before the first word is spoken. You may ask: how does the network originate? I cannot answer this question except to say that it may be a chicken-egg type empty question.

The novel contribution I wish to make to philosophy of language is to extend Searle's cluster theory of meaning to descriptors that indicate the ontological category of beings. Quite simply, it is part of the meaning of a word whether what it denotes exists or not. Consider: when Jill tells Jack that she has seen a unicorn in her back garden, the reason that Jack may not believe her is that if Jack is like most people he regards unicorns as mythical. The descriptor "mythical animal" is one most people associate with the word "unicorn". Why? Because we learn this almost from the time we learn the word. Similarly, along with descriptors like "chief of the Greek gods", and "father of Hercules", a descriptor we associate with the word "Zeus" is "non-existent". Why? Because we learn that the word "Zeus" is a figure in Greek mythology and when we first encounter Greek mythology we also assimilate the belief that all the beings in Greek mythology don't exist and never existed (although we may later learn that the city of Troy did in fact once exist). Descriptors associated with "Sherlock Holmes" include "greatest detective in nineteenth century London", "close friend and confidant of Dr Watson" and "fictional character created by Conan Doyle". Thus some of the descriptors associated with "Sherlock Holmes" seem to characterise him as real and others do not. The fact that people can disagree about whether some objects, like God for instance, exist or not strongly supports the hypothesis that whether something exists or not is a part of the meaning of the term used to denote it and can vary among people. Once we recognise that ontological status, existence or non-existence, can be part of the meaning of words, this leads us to the conclusion that Searle's cluster theory, although a good start, is not a fully adequate theory of meaning. Its inadequacy is that Searle's theory seems to imply that the meaning of a proper name is a list of descriptors. But if want to explain how the meaning of "Zeus" can be attached to both "wielder of thunderbolts" and "non-existent" we need to recognise that the meaning of a word must also involve very complex relationships between different descriptors. The meaning of a word is a structure of which ontological category might be a kind of base-level component. And, to reiterate, this structure is not entirely verbal and is not wholly isolable from the structures of other related words.

I'll make an observation concerning this, something I find fascinating. Children learn that some things are real and others fictional almost from the time people start telling them stories. It seems that the imbibing of fairy stories and fantastical narratives, like the Beatrix Potter tales and the Harry Potter books, is an important part of child development. Children need to be exposed to fantastical, literally false, stories, and they seem to enjoy fictions precisely because they are fictions. No culture in the world thinks that children should learn to read by being forced to study encyclopaedias. To have a robust sense of reality we also seem to need a sense of the unreal. Perhaps child psychologists have delved into this interesting topic but I do not know if there is ever been any philosophical exploration of it.

The meaning of 'meaning' is closely related to the meaning of 'definition'. It can sometimes seem that these two words are identical but I shall suggest that there is an important nuance. Today we often think that the definition of a word is what we find in good dictionaries but there is a problem with this view – as Quine pointed out in his famous essay "Two Dogmas of Empiricism", lexicographers ultimately determine the definitions of words by closely studying how people in their linguistic community actually use them and, when there is uncertainty, adjudicate between different meanings by kowtowing to the usage of famous writers, like say Milton, or experts, rather than that of ordinary yobs in the street. The aim is to find some balance between how people generally use words and how they ought to use words. The definition of a word is shorthand for the meaning most people generally ascribe to it or the meaning they ought to ascribe to it should they accept the opinions of authorities. When we recognise this, it leads us to a total reconceptualisation of much philosophy. Philosophers and others throughout history have often argued that some claims are true by virtue of the definitions of the words contained in them. The sentence "A bachelor is an unmarried man" is true by definition, supposedly. "God exists" is true by definition, supposedly – this last is an important version of the ontological argument. Years ago a psychiatrist said to me, "It is impossible to recover from schizophrenia by definition." With an improved understanding of definitions, we can reinterpret claims such as these as saying, for instance, "God exists because everyone believes He does and this is the view of authorities on the matter". Now, in the Middle Ages, the descriptor "existent" was almost universally attached to the word "God" and this belief was strongly encouraged by church authorities. But today believers may be in the minority and this descriptor may no longer, for most people, be part of the meaning of the word "God"; thus it is no longer possible to say "God exists by definition." In my dictionary, a contemporary dictionary of course, the main definition for "God" runs as follows: "(in Christianity and other monotheistic religions) the creator and ruler of the universe and source of all moral authority; the supreme being". Note that this definition includes an ontological category: "in Christianity and other monotheistic religions". It is up to the person consulting the dictionary whether or not she wants to believe that Christianity is true. We could imagine a definition of "Sherlock Holmes" as being "(in the novels of Conan Doyle) the greatest detective of nineteenth century London." This line of reasoning suggests that an argument of the form "Sentence A is true by virtue of the definition" is not conclusive because it relies on the infallibility of the beliefs of the linguistic community who have together determined the meanings of the words that make up the sentence. It is not conclusive because the linguistic community might be wrong.

The perspective which I am presenting in this essay is based on a view of language as something that mediates between individuals in the form of conversational interactions. Even when someone writes an essay, as I am doing now, I am seeking to communicate my ideas to others, hopefully to you, the reader, although I have very little notion of who you are. Now, it is certainly the case that a person may maintain a kind of internal monologue in his or her head, a monologue in which the originator of the monologue and its audience are the same person; Chomsky has argued that language evolved principally to be the building blocks of thought, with communication as a secondary, derived function. But I myself still adhere to the old-fashioned view that communication is the primary purpose of language and that its role in private cognition is secondary, although I am not absolutely committed to this view. Conversational interactions require that there be considerable similarity between the semantic structures of speaker and auditor but would be redundant if the two were absolute identical. My perspective is that we have to regard language users as all individuals interacting verbally with each other. This perspective runs counter to the general paradigm of much philosophy, especially in the anglophone tradition. There is a tendency to regard the linguistic community as monolithic: "we" think this and "we" think that. Kripke's theory of proper names, for instance, assumes the unitarity and eternality of human civilisation implicitly : he supposes that two and a half thousand years ago, Aristole's parents baptised him "Aristotle" and that there has been an unbroken causal chain linking speakers today with that original baptism. Kripke does not discuss the links in the causal chain (the conversational interactions), does not account for how a word like "Zeus" (which lacks a referent and therefore never involved any kind of baptism) can be meaningful, and has trouble accommodating the fact that different individuals can have different beliefs and understand the meanings of words differently. Although I concede that there is very significant overlap between the semantic structures of individuals within a linguistic community, that it is not altogether disingenuous to say "we" think this and "we" think that, I am highlighting the differences between people. And, to repeat my most important point, although not all language use involves the communication of information between individuals, this is a major role of language, something that would be redundant if there were no differences between the mental networks of different individuals.

Finally we come to the issue of Truth, something that has been looming over the discussion. The best way to approach this question is through some case studies. In the early twentieth century, physicists believed the universe as a whole to be eternal and unchanging. The reason Einstein included the Cosmological Constant in his theory of General Relativity was because, without it, his equations predicted that the universe would be either expanding or contracting, and he did not believe this to be possible; Einstein later called this inclusion his greatest blunder. It was with the discovery of the red-shift of distant galaxies by Hubble that physicists concluded that the universe was expanding and, together with the discovery of the cosmic background radiation, this led to the postulation of the Big Bang. It is tempting to think that the physics community changed their minds en masse overnight but of course this is not what happened. Physicists generally believed the universe to be eternal and unchanging in 1915, then subsequently Hubble and others found empirical evidence for the expansion of the universe, came to the conclusion that the universe was expanding, and then persuaded others of this 'truth'. This 'truth' propagated out to others, through 'conversational interactions', but this did not happen instantaneously – in fact, the physicist who coined the term "Big Bang" did so to show his disdain for the theory. He chose a ridiculous term for the theory because he thought the theory was ridiculous. Gradually the theory of the Big Bang gained acceptance throughout the physics community, through persuasion, persuasion usually employing corroborating observational evidence that physicists accepted, and from there disseminated out to the lay public. Even today, however, people can be found who think the universe was created by God in 4004BC and the physicist Roger Penrose has argued, if I recall correctly, that some stars predate the Big Bang, are remnants of a previous universe, and that therefore the Big Bang was not the ultimate beginning . Thus what we think of 'truth' must be considered a kind of consensus opinion created through persuasion and justified by one's faith in authorities, in experts, and even then dissent can exist. In the case of the Big Bang, the success of this theory is partly attributable to people's faith in experiments and empirical evidence as reported by credible scientists, well replicated experiments and observations, faith generally in the scientific method, as objective support for the arguments made by theorists.

Consider now the term 'schizophrenia'. The situation with this word is very different than that with the Big Bang because there is much more obvious disagreement about how to define this word than there is disagreement today about whether or not the Big Bang occurred. The most authoritative definition of this word might be that found in the DSM but still there are great divergences of opinion about its meaning because the word 'schizophrenia' is I believe somehow shapeless, can take on any meaning anyone wants to associate with it. Robert Sapolsky, in Determined, regurgitates the faddish new definition of schizophrenia as "a disease of disordered thought", but recently I have been reading The One Thing We've Never Spoken About by Elfy Scott, a book which does not attempt to define 'schizophrenia' but nevertheless presents a picture of this condition, a meaning for this word, quite different to that presented by Sapolsky. Scott is the daughter of a schizophrenic mother, has studied it, and has spoken to many people who have been diagnosed with it; I think her view of 'schizophrenia' is vastly superior to Sapolsky's view and that Sapolsky had waded into territory he knows almost nothing about. Then we might consider this blog in which I have been discussing the meaning of 'schizophrenia' for many years and presenting a different view again. It has occurred to me recently how problematic the definition of schizophrenia is because, although I have often experienced profound mental distress over the last ten years, this distress didn't really take the form of hallucinations or delusions but rather, for instance, of terrible nightmares, and was a reaction to my treatment by the Mental Health Service; nevertheless I was officially diagnosed 'schizophrenic'. It is not the purpose of this essay to elaborate on this self-reflection. The point here is that the meaning of the word 'schizophrenic' varies widely among people, that there is no consensus opinion with respect to its meaning or with respect to what we can say truly about it. With respect to 'schizophrenia', our notions of truth break down. Truth here is a matter of negotiation between people about how best a term like 'schizophrenia' should be defined and it might be that the word 'schizophrenic' is so compromised that we should discard it altogether.

In the post "Rationality and Irrationality", I discussed how in anglophone philosophy there is confusion between the descriptive and the normative. Philosophers want to stipulate how we ought to determine whether a sentence is true or not whereas I want to describe how people actually decide in the real world whether some sentence is true or not. We need to recognise that much of our 'knowledge' about the word is communicated to us through language, through 'conversational interactions' if we broaden this term to include linguistic artefacts such as news stories, scholarly essays, books, and so on. Everything I know about Aristotle, I have either read or been told by someone else. When someone communicates to someone else information new to him or her, the receiver will tend to believe it, to regard it as 'true'. If different conversational interactions convey conflicting information, we make decisions about what to believe based on prior knowledge, on reasoning processes, and often by deciding which of the communicators we feel should be regarded as most credible. Where there are differences of opinion within the linguistic community people seek to persuade others and if we define 'rhetoric' as 'the art of persuasive speech' we must consider rhetoric to be a core part of language use. I strongly believe that much of a person's beliefs are formed by the person conforming his or her beliefs to those expressed or espoused by those he or she finds trustworthy, an idea we find in Kenneth Burke. We could throw the notion of 'truth' away completely but I don't want to do this. Rather I would like to suggest here that any notion of truth must depend on a kind of faith. Generally we have faith in those around us to tell the truth and when there are differences of opinion we choose who to have faith in. Although I have not observed the galactic red-shift myself, I put my faith in the physics community, as represented by someone like Sabine Hossenfelder, in choosing to accept their consensus opinion that the universe is expanding. I put my faith in Michael Cohen over Donald Trump with respect to whether or not Trump fiddled the books. I put my faith in Steve Braunias when he says that Mark Lundy was actually innocent of the crime for which he was convicted, the murder of his family. Of course one's faith in someone else may always be misplaced but we have no other option.

I'll conclude the essay by summarising the theory of meaning I have proposed. The meaning of a word involves a reticulated network of associated words, sensory memories, and perhaps concepts; the meaning of any one word cannot be isolated from the meanings of other words. Knowledge of language is learnt and much of the way it is established is by means of 'conversational interactions'. Conversational interactions rely on the participants possessing very similar meaning networks but there would be no point speaking or talking to anyone else at all if the participants all possessed exactly the same linguistic networks as each other; this is the reason conversations occur at all. Through conversational interactions we align our networks with those around us and with those we trust. When there are differences of opinion within the linguistic community about what is true we make decisions about who to believe based on principles like consistency and the apparent trustworthiness of those speaking with us. We place faith in those around us to tell the truth. This theory may seem cynical: we may contrast it with the position of someone like John Rawls who argued that the principle of universal human rights follows directly from the "definition of justice". Just because it is cynical, however, does not make it incorrect.

This essay may not be that interesting to many of my readers and I am not sure if I have presented my theory as clearly as I could. However this topic is interesting to me and maybe one or two readers out there will find it thought provoking.