Language does not "convey thoughts"

I am reprinting here a recent Substack article of mine in which I try to articulate in a very concise way some ideas and intuitions I have been developing about language and communication.

Some months ago I mentioned that, after having spent some time in countries where proficiency in English is not widespread, it was a relief to be in a country which presented no linguistic challenges for me. My focus in England has been on acclimatizing myself to English ways. To say “reacclimatizing” would probably be misleading, given how long it is since I’ve been here and how much things have changed in the intervening decades. I may write about shifts in the social and political mood and changes in the wider culture at a later date but, for now at least, I want to steer clear of politics and related matters.

Lately I’ve been reviewing my views on language, thought and communication, and what follows is a brief statement of my general perspective on the relation between language and thought and the implications of such a view for human communication. Even when there is no “language barrier” in the usual sense of that phrase, effective communication is far from guaranteed. In fact a shared language will often contribute to creating an illusion of agreement, obscuring profound differences in point of view.

In a footnote to one of his essays, Aldous Huxley (obviously recalling a personal experience) talked about an unbridgeable gulf suddenly opening up between two men engaged in a friendly fireside chat on account of a stray remark. A felt or imagined affinity proved to be entirely illusory.

My point is that language, by its very nature as a cultural phenomenon, tends to create illusions of affinity. Its nature is not so much to reveal as to conceal or paper over the very real differences in how speakers see the world.

Language, of course, is a powerful tool for doing what it does best. It facilitates thought and interaction and so makes other aspects of human culture possible. But it doesn’t operate in the way we naively think it does or deliver exactly what it seems to deliver. What linguistic communication doesn’t do is convey thoughts in a literal or even (I suggest) in a metaphorical sense — though we often fool ourselves into thinking that it does.

Let me explain. I have a thought or feeling. I try to put it into words. If it is a thought which draws not only on my commonsense or technical or scientific knowledge of the world but also on my personal memories, values or judgements then it is private to me: it cannot be encapsulated in a string of words and so cannot be conveyed into someone else’s brain. The words, the sentences are in some sense conveyed, sure, but the crucial point is that the same words and phrases trigger different thoughts and feelings in different brains.

The basic building blocks of a language are phonemes (minimal, meaning-distinguishing sounds). Every language has its own set of phonemes from which words and phrases and sentences are built. In any given language the abstract forms of words (lexemes or word stems) constitute a dynamic lexicon which is involved in complex combinatorial processes involving the formation of actual words and sentences (processes studied under the rubrics of morphology and syntax).

At least a rough distinction can be made between language-specific brain areas and non-language-specific brain areas. Phonological and morphosyntactic processes operate largely unconsciously in specific areas of the brain and are relatively self-contained. At a certain level of abstraction, these processes can be explicitly described and modelled (i.e. formalized). But semantic processing spreads a much wider net in terms of the brain areas involved and is not formalizable. It draws not only on language-specific brain areas but also on non-language-specific areas, including those dealing with emotions, memories of lived experience, general reasoning and problem-solving, etc.. This point has been dramatically illustrated in fMRI studies.

The key point here is that semantic processing — unlike phonological and morphosyntactic processing — draws on general aspects of thought, including memories and patterns of emotional response which are necessarily unique to the individual. The complex phonological and morphosyntactic processes which give human language its remarkable power are, in an essential sense, shared. These processes only arise and sustain themselves over time within a shared social and cultural milieu. They represent something that speakers of a given language have in common. The semantic aspect of language also involves shared understandings of course, but the sharing involved here is of a much more limited nature.

Take a simple noun like “dog”. It has a publicly-agreed primary meaning or denotation and various other (also publicly-agreed) senses. But even when used in its primary sense to refer to a particular kind of animal, the thoughts and feelings and memories triggered in my mind by this word will (because of our different personal histories) be very different from the thoughts and feelings and memories which the word triggers in your mind. And if this is so with such a mundane word as “dog”, how much more divergent will thoughts triggered by more value-laden or abstract expressions be?

Such differences relate to (but cannot be completely reduced to) the concept of connotation. Connotations are, like denotations, usually publicly shared (at least to some extent) but, being tied to emotion and personal memories, they are necessarily more vague and difficult to define. Certain words are universally accepted as having a negative connotation (like the noun “lackey”, for example, or the adjective “sanctimonious”) but in many cases the connotation is not obvious or easy to describe and different speakers may report diametrically opposed attitudes.

Though the brain processes underlying linguistic processing remain obscure, the basic elements and patterns of a language — apart from the intricacies of semantics — can be discovered and written down. Linguists have been doing this sort of thing for hundreds of years. In principle at least, anyone with access to a dictionary and a grammar can attain technical proficiency in a new language. All of this suggests that the core mechanisms of language are neither deep nor private in any meaningful sense.

Our personal thoughts, by contrast, remain essentially private, however much we attempt to put them into words or express them in other ways. Literary and other artistic creations are notoriously subject to multiple, divergent interpretations. Gestures can be subtle and expressive but are limited in scope and range.

In some ways, the idea that our personal thoughts remain forever private is confronting. But there is also comfort to be found here, especially in an age awash with spin and propaganda and dominated by social media. For there is, after all, a sacrosanct space where judgements are made or withheld, and where integrity and truth still matter and cannot be threatened.

In fact, the idea I am trying to convey here (an idea, note, not a thought!) can be extremely valuable in helping to minimize frustration and stress.

And lowering our expectations concerning what language can reasonably be expected to do will also (somewhat paradoxically perhaps) make us better communicators.