Blog Post: Prescriptivism and Descriptivism in Language

I thought I would write something about a phrase I see a lot in online discussions about language. I’ll throw this up on Substack once I’ve had a chance to polish it, but I thought you might enjoy it in its very much unpolished form :).

————-

It has become a kind of thought-terminating cliche in discussions around language to say something like “definitions are not prescriptive, just descriptive”. As it happens, this is a statement I mostly agree with, in that our definitions of words track their usage. There is no Platonic form of the word “dog” which describes its definition once and for all. Language does evolve over time.

However, it does not follow from this that definitions are never prescriptive. After all, imagine I said the following sentence to you:

“Dogs bark merrily in the moonlight”

But by the word “dog”, I in fact meant “cat”, and by “bark merrily” I mean “hunt”. If there was no prescriptive component to language, there would be nothing wrong with this. I would simply be using language in a different way, no better or worse than your own. In one sense this is true, in that I have not committed some moral ill. But not all normativity, or prescription, is moral in nature. We acknowledge that there is a certain normativity in language when we correct young children on how they speak. My little cousin, when she was younger, used to point at any man around her father’s age (including men on the telly), and say “dada”, at which point we would say something like “that’s not dada, that’s John Malkovich” and generally all be delighted by how adorable this linguistic mistake was. It was, however, a linguistic mistake. It was a misuse of the word “dada” to refer to someone who was not, in fact, dada. In order to understand why definitions, word usages, and language more generally is not merely descriptive, we have to answer the following question: why did we correct my little cousin, and why did we recognise that she was using the word “dada” incorrectly?

Here is one potential answer: when someone speaks, we presume that they want to be understood. It is our default interpretation of their aims. Moreover, I think we consider it a general good that language facilitates the accurate interpretation of speakers’ intentions. If I had a language where there was no consistency in usage, I simply would not be able to communicate using it. I would not be able to trust that when I say a given word, that word would be an accurate representation of the thought I am trying to inspire in the head of my audience. It is easy to think this is a deeper point than it actually is. I am just making the following (rather banal) observation: part of the general use and function of language is to ensure speakers are understood by hearers. In order to do this, you need some kind of reliable lexicon. Thus, a reliable lexicon is a relatively core component of a language achieving the goal we have (collectively) assigned to it. A language with a reliable lexicon is a “good” language in the same minimal way that a sharp knife is a “good” knife - it is good for the thing we want to use it for.

It is this component of language that facilitates a certain degree of normativity. It is almost a foundational principle of pragmatics (the field of linguistics that deals with language as we use it) that when we interpret a speaker, we assume that they wish to be understood. This is part of how the philosopher and linguist Paul Grice thought we detected implications, but you can find it in some way, shape, or form, in many modern areas of pragmatics (e.g., Relevance Theory, Neo-Gricean pragmatics). To see why this is, we can only reflect on how our interpretive apparatus would work without this assumption. If we did not go into our conversations with the general idea that the people we were listening to wanted to be understood by us, then we would have no basis on which to even get an interpretation of their statements. If we don’t think they are even trying to communicate, then we cannot glean anything from their statements in the first place.

So when we try to correct my little cousin’s usage of the word “dada”, what are we trying to do? I would say something like the following underlying thought process is at play (understood as a rational reconstruction - I’m not suggesting we had these exact thoughts at the time):

We assume that she wants to communicate her thoughts accurately to the people around us

We know that the word “dada” will be understood by her listeners to refer to her father, and not other men around her father’s age

Thus, her usage of the word “dada” is out of step with how people will interpret her words

Because we love her, we want to ensure that she achieves her goal of being understood

Thus, we correct her usage of the word “dada”, hoping that the 37th time we have done this, she will cotton on.

So from the initial assumption that she wants to communicate a thought accurately, we have arrived at the rationale to correct her usage of the word. If language (and more specifically definitions) were purely descriptive, we would have no basis upon which to do this. But since we assume an aim to be understood, the whole process works smoothly, and basically unmysteriously.

To take this back to our original point, this is why I have trouble with the statement “definitions are not prescriptive, just descriptive”. It only tells half the story. Under the assumption that someone wishes to be understood, definitions are both. The common usage of the word forms the backdrop against which we judge which usages of a word are most likely to be understood, and the assumption provides the minimal amount of normativity/prescriptivism needed to separate clear “correct” cases of linguistic use, from “incorrect” ones.

The flipside, of course, is when someone pretends that the original usage of a word is set in stone forever, never to change. This ignores the descriptive basis of definitions we just talked about. The basis of someone saying, for instance, that the term “literally” should never be used to mean “figuratively”, is the proposition that the use of “literally” in this way is likely to lead to confusion. As soon as this stops being the case, it is fair to say that the definition of the word has changed. At one point, the literalists about “literally” were probably unquestionably correct. They certainly would have been the first time someone used this linguistic innovation. But slowly, popular usage has shifted, and the definition along with it. At least within certain linguistic communities, it would no longer be confusing to use “literally” in this looser sense. I think this is the general argument appealed to when someone says “definitions are not prescriptive, just descriptive”. But the idea that definitions are set in stone for all time is only one type of prescriptivism or normativity about language. The more mundane kinds of normativity are often ignored. This is hardly the end of the world, but I do think it gives us a misleading picture of language, and that is a bit of a shame.