Theory of Descriptions

Reference, Truth, and Meaning

How is it that language can pick things out in the world, and at the same time be meaningful? Throughout the history of philosophy, there has been a view which attempts to answer such questions by supposing that words attach to reality by adopting things as their meanings; more specifically, that the ontology of language and the ontology of the world are identical: subjects of language are subjects of reality, and the predicates of language are the predications of reality; for example, the term “William James” picks out the one and only notorious psychologist from the 1900’s that worked at Harvard and wrote the “The Principles of Psychology” vol. 1 & 2. The name “William James” picks out a specific object and has direct reference to that object. In addition to proper names that denote a person, there are also proper names that denote landmarks: I.e., the term “The Empire State Building” references a building within New York that has N-floors and is a well-known tourist site. As we can see, the meaning of a word is the ontological agent or property which it picks out.

By the logic of the above-explained doctrine, sentences only have reference when they pick out their corresponding object. No discrepancies between the label and the object can be tolerated. In example: “A wolk jumped the tree”. Because the term “wolk” has no reference, it follows that the entire sentence is void of meaning. On top of sentences being viewed as meaningful or un-meaningful, sentences can also be viewed as either true or false with truth-conditional semantics; only because words have conditions for their reference, and a failure to meet the conditions leads to a falsehood. For instance, “there are at least ten, but no more than ten million, words within this paper,” is true so long as the conditions of the proposition are met. And given that we are short 9.9 million or so words, such a sentence must be false.

Thus, true sentences must be meaningful whereas false sentences must be void of meaning. These two entailments of the reference theory of meaning give us a neat and tidy picture for words, but there are some problems.

We now arrive at a paradox: namely, that there are sentences which we perceive as meaningful yet have no reference. If truth-conditional semantics requires conditions to deem whether something is true or not, then words which lack reference cannot be subjected to truth-conditional semantics given that they lack conditions of satisfaction. Thus, the notions that meaningful things are true things and that un-meaningful things are untrue things, seemingly, produces a paradox. For example, Santa Claus is an elf is meaningful, yet lacks direct reference to reality, and so cannot be assessed for truth-conditions. How does Santa have reference when Santa does not exist?                                  

Criticism of Theory of Descriptions

There are plenty of issues with this theory: to name a few, it takes a literalist approach to language; it defines existence in a remarkably arbitrary fashion, which excludes things that do exist; and it commits its self to platonic realism by supposing that words have reference. Let’s discuss these points further below.

On Literalism and Logic 1.1.1

Bertrand’s approach to language is perhaps one of the most literalist approaches to language, particularly because of his favouritism of the application of logic, and it is the paragon theory of how a strict logical analysis of linguistic content fails to explain language. Of course, there is nothing a priori wrong with logic being applied to language investigations, though I argue that logic needs to be applied on the level of language production mechanisms as well, and not the content of the language alone. Bertrand’s approach deals with the content of language only: meaning, the relations between words and reality. Such an approach is similar to checking the end score of a sports event rather than watching the sports event: one only perceives the result (score) rather than the mechanisms (players) responsible for the result.

The problem with this content and logic approach is that the content of a language lacks all the necessary truth-conditions, which means the logic is dealing with only a fraction of truth-conditions associated with language; and in addition to that, symbolic logic is neither probabilistic nor capable of producing meaningful contradictions, both of which are necessary for an understanding of language. Essentially, Bertrand was headed in the right direction but missed a lot of the details.

Logic, Language, and Truth-Conditions 1.1.2

One important condition to have information about is the brain; more specifically, what neural activity occurs in relation to what linguistic activity; for instance, semantic processing has often been argued, and demonstrated, to occur around the brocas area, among other places (Pinker, 1995); this insight has helped us understand both language and language disorders, and so insight into the brain, therefore, helps us understand language. There are also other reasons, other than insight into the semantic nature of language, for why data on the brain is necessary, one being meta-representations.

Suppose a man is suffering from hallucinations, and that his hallucination is of Santa Claus. If this man screams, “I see Santa Claus!”, then he would have said something false because the conditions cannot be found within the environment – the reference for Santa would require different empirical conditions for the utterance to be true. However, what the man is screaming indeed refers, which is different from reference, to something true, though what is true is that his utterance reflects a meta-representation; meaning, he is attempting to convey information about his representation of Santa Claus by discussing the representation; much like someone who is discussing the beliefs they have about their own beliefs. These meta-conditions cannot be found within language because language cannot make a discrepancy between the two styles of representation: meta and non-meta; not only does language fail to discriminate between these two styles or representation, but also Bertrand’s theory of description fails to discriminate between the two modes. Thus, a logic which only looks at the language its self will be unable to discriminate between “I see Santa Claus!” when uttered by someone that is hallucinating, and ‘I see Santa Claus!” when uttered by a child who is at the local mall during Christmas.

However, a logic which assesses the frequency at which a cognitive function occurs in relation to some linguistic categorical variable, indeed, has a much better chance at understanding language. Simple binary models applied to the content of language alone are woefully insufficient. Thus, knowledge about the conditions of the brain is necessary for a logic of language, given that the brain, and not the language, performs the meta-cognition.

In addition to meta-representations, a truth-conditional approach to language will also require information about brain functions: specifically, those associated with and responsible for learning and conditioning, given that those mechanisms introduce new items into the lexical set. Put more explicitly, there are neural mechanisms which interact with the symbols and sounds within the environment, and these mechanisms are subject to Pavlovian conditioning; thus, the way these mechanisms interact with sounds and symbols is subject to change, and this change can be triggered by variables within the environment that may go beyond our comprehension. So, a fundamental understanding of the brain is necessary if we are to apply logic to language; only because the content of language is not finite. Each new item introduces new truth-conditions, assuming words have reference, and so a symbolic logic of language would require a seemingly infinite amount of truth-conditions to deem all the words true or false – and no living thing can compute all truth-conditions. However, an understanding of the of mechanisms necessary for learning and language production may allow us to formulate abstract rules as the truth-conditions, as opposed to the conditions which a word references. For example, if I utter “ajk” and point at an object, human listeners can use their Default Neural Network (DNN) – the neural basis of theory of mind (Lieberman, 2014) – to understand my act of denoting; that is, the listeners can infer that by “ajk” I mean the specific object at which I pointed. And in addition to that, over time, I could simply stop pointing and people would still comprehend the utterance “ajk” –  so long as they have made the association between the object and the utterance beforehand. From there on, one can begin to formulate abstract rules to represent possible conditions of satisfaction. And here, the truth-conditions would indeed be physical rather than platonic or introspectively determined – more on this in the platonic realism section.

Of course, the brain is not all that needs to be mapped, though it is a piece of the puzzle. Indeed, cultural context, social context, and so forth can all influence language while maintaining the same neural functions; consider, emotion physiology is quite rigid in its variability across the human species, and yet different cultures express emotions radically different due to cultural influence (Griffiths, 1997). By the same token, we can infer that the neural mechanisms for language can be identical in structure but interact with the environment differently, or perhaps have different content stored within the memory systems. At any rate, there is a lot of work to be done for a truth-conditional theory of language, so much work that one might consider more probabilistic approaches instead.

Meaningful Contradictions 1.1.3

The law of non-contradiction is certainly correct when it comes to the physical nature of reality, but it is not very useful when it is applied to the content of language; for instance, it is common for people to say, “don’t worry about me, go hang out with your friends,” while they really mean, “worry about me, and don’t hang out with your friends”. Such a conundrum is an example of a meaningful contradiction; which is to say, it is not pure nonsense even though it is contradictory. This is because the neural conditions are different from the linguistic conditions; and so, the law of non-contradiction would lead us to falsehoods if we applied it to the content of language alone. It is when we look to the other conditions of language that we see meaningfulness in a surface contradiction. So, a coherent Truth-Conditional theory of language must be able to account for meaningful contradictions, which can only be done by cross-examination of various conditions associated with a given phenomenon: I.e., neural conditions compared to linguistic conditions.

Literalism & Metaphor 1.1.4

Lastly, when applied to language, logic assesses the conditions of satisfaction for a true or false claim; and so, one cannot speak in a non-literal fashion without stating a falsehood. For example, anyone who says, “time is money,” or, “love is a battle,” will have said something false, which means Bertrand’s approach excludes metaphor. This ought to raise suspicions given that both myself and Bertrand make use of metaphors throughout our papers, and most people use metaphors in their daily conversations as well (Lakoff, 1981). However, if we bring neural conditions into our picture, then it is at least foreseeable as to how one can overcome this difficulty. That is, suppose there is a neural condition for the semantic memory associated with “battle,” and that this semantic memory had multiple nuanced meanings stored away, one of which was loosely associated with “challenge,” then it becomes feasible to apply truth-conditional semantics to our examples: namely, “love is a battle” can be interpreted as “love is a challenge” because the neural condition associated with challenge was activated upon the utterance of “love is a battle”. So, there are N amount of associated meanings for “battle”, and neural condition X can indicate one of those N amount of associated meanings which get selected by either the speaker or listener; thus, solving the problem that logic has with metaphor.

Not only does Russel’s theory exclude metaphor, but it would also exclude knowledge about fictitious characters. In Russel’s view, to say that Santa delivers presents once a year, undoubtedly, is a false statement. This brings us to his arbitrary grounds for the claim of existence.

Arbitrary Grounds of Existence1.2.1

For those who have read Bertrand’s works, then it should be no surprise that he denies the existence of things like “Santa”. This part of his work often confused me; to clarify the nature of my confusion, consider the following: the method which we use to declare the existence of “Santa” is identical to the method we use to declare the existence of a country; more specifically, since we cannot point to lines on land to denote a country, then we must point to lines on a piece of paper; and so, a countries borders, and therefore existence, depends on the lines we choose to draw on paper. Now, if one wishes to argue that countries are not determined by boundaries (lines), then they will have to demonstrate such without creating boundaries. Evidently, the only way to suggest that Canada is different from America topologically, and that two are more than a single continuous landmass, is to demarcate one from the other based on cognitively constructed borders (I say cognitively constructed because the borders are a projection onto the environment; they cannot be found in a user-independent world). Moreover, this process is equivalent when it comes to the existence of Santa Claus; the existence of Santa Claus depends upon lines drawn on a piece of paper, albeit lines that are far more dynamic. So, even though these two entities are epistemologically equivalent, Bertrand would only accept one as legitimate; hence my confusion. And in addition to that, both Santa and Canada, though drawn on paper, are perceived through the senses, and so now Bertrand must say that we can perceive things which do not exist. For instance, if I have a drawing of an object which I declare is Santa and a drawing which I declare is Canada, and I then utter sentences about either drawing: for example, “Santa delivers presents,” and, “Canada is where I live,” only the utterance about Santa is false because Santa, supposedly, does not exist. If that is to be the case, then what was I referring to?

Now, there may be those who will argue that the lines on the map have reference to a landmass, whereas the drawing of Santa has no reference; however, such a charge would be based on a misunderstanding of the above-mentioned argument, as well as a form platonic realism. On the one hand, it is a misunderstanding because my words would have been a about the contents of the paper for both Santa and the map, not physical geography. So, to suggest that the map of Canada is legitimate because it has reference to something else is to say that one cannot have reference to either the drawing of Santa or the map. If I am talking strictly about the map and the drawing, then it matters very little if either the drawing or the map reference something else. On the other hand, to suggest that the map has reference to a physical landmass is platonic realism, though we will see why that is the case further below.

On Reference and Platonic Realism1.3.1

The notion of words having reference assumes that the words have a non-empirical property; only because reference is a notion that suggests a relationship between a word and the physical environment without indication of a mechanism for interaction; and thus, one must conclude that the notion of reference is platonic. Comparatively, a notion of referring is not platonic because referring is something done by a brain; and so, someone who states that the word “Santa” has a common referent (not reference) is suggesting that a collective of brains commonly refers to a specific object with the word “Santa”. Reference has no mechanisms for physical interaction, and referring relies on a human brain; therefore, reference is platonic and referring is empirical in nature.

Now, being a platonic realist is entirely fine, though I assume its inherently wrong given that it cannot produce empirical results; which is to say, reductionism and physicalism have brought about far more tangible consequences than Platonism; and as such, on the grounds of practicality, I side with physicalism.

Conclusion 1.4.1

Bertrand’s theory is too shallow in its analysis; indeed, a truth-conditional semantics of natural language requires more conditions than the content of language alone: I.e. environment, neural, etc.., In addition to that, Bertrand’s theory also commits its self to platonic realism via reference. Reference notions of word meaning are platonic because they fail to indicate a physical mechanism between a given word and the environment; that is, reference suggests words have an inherent property that attaches the word to a physical environment, whereas referring suggests that the brain learned that certain words refer are associated with certain objects. And lastly, Bertrand had had a remarkably arbitrary way of declaring existence, given that his method suggests that one drawing is more legitimate than another, even though they are identical in form; his view on things that do and do not exist commits him to a few absurd claims which are, in my view, not tenable. Thus, Bertrand was ultimately wrong in his theory of Descriptions.

References

Lieberman, Matthew D. Social: Why Our Brains Are Wired to Connect. , 2013. Print.

Lakoff, George, and Mark Johnson. Metaphors We Live by. Chicago: University of Chicago Press, 1980. Print.

Lycan, W. G. (2000). Philosophy of language: A contemporary introduction. London: Routledge

Martinich, A. (2008). The philosophy of language. New York: Oxford University Press.

Pfeiffer, G. A. Russell’s Introduction to Mathematical Philosophy. Bull. Amer. Math. Soc. 27 (1920), no. 2, 81–90.

Pinker, Steven. The Language Instinct. New York: HarperPerennial, 1995. Print

Griffiths, Paul. What Emotions Really Are: The Problem of Psychological Categories. Chicago, Ill: University of Chicago Press, 1997. Internet resource.