On misunderstanding

The grades I remember most often receiving in school were “A” and “Incomplete.” Focused on meeting some arbitrary idea of greatness, I would pull back from the game, try to ascend above it, and get lost in self-critique — to the point where I’d fail to deliver on the original goal.

Let’s suppose it’s possible to define myself, to put words to that unfillable hole at the center of sensory experience. I am a creator because I create things. I have called myself a designer because I have spent considerable time perfecting the appearance of shapes, lines, and color in two dimensions. I have called myself a writer because I have learned about the intricacies of syntax and the importance of emotion and reason in the written word. I have called myself a programmer, because I have the most basic ability to construct messages in such a way that they are understood and executed by machines for some business goal. But I’ve been questioning whether that last title accurately describes whatever it is I want to accomplish in this life.

One piece of advice I received lately, from a mentor who was reviewing my code, was this: your job as a programmer is not just to write code that passes through the linter and compiler. The goal is rather more abstract; that is, to do something useful, to combine data such that it is legible to my colleagues and illustrates an underlying pattern which the consumer may not have considered. I hadn’t been doing that. Rightfully so, I was called out for behaving like a suck-up reporter whose only goal was to get his piece published or get his name on the masthead.

After all, the work of writing code is just a highly specialized form of writing. Naturally I had set up adversarial relationships to the compiler, the code reviewer, the interpreter. I had decided the bulk of the burden rested on them to interpret my work and not I to communicate its meaning.

The trouble is that I have been once again wrestling with the concept of meaning. I have been waging an internal struggle for which there is perhaps no solution. Meaning is felt, but it is also expressed. When a speaker says (expresses) something unclear, the response is usually, “Can you clarify what you mean?” to which the speaker’s response is either “Yes, let me rephrase that...” (re-expression) or a situational excuse like “Not at this time, let’s move on and it will become clear...” (silence, lack of expression).

A speaker’s reaction to the problem of being misunderstood therefore falls into one of two categories: re-expression and silence. Some choices made in diction can be stripped away, without changing the position of the statement on this axis of expressivity (its condition as an engaged response or a null response).

The expressive response is still expressive if the response is the same as the first statement, that is, repeating oneself word-for-word can sometimes constitute all the clarification that may be needed on behalf of the recipient. On the other hand, a verbose version of the null response has the same semantic meaning as a less verbose one in relation to the first statement. For instance, “I will not repeat myself because I must get going; would you look at the time...” provides no more or less semantic information than “No.”

Yet these reactions would not likely be received in the same way. Why?

We should consider there to be another axis of emotional salience, which is perpendicular to expressivity and can serve to explain the difference in tone. Emotional salience can be crudely split into positive and negative sides, with degrees. Such a model can differentiate between “Let’s not do that” and “I really, really, would hate to do that,” the second statement being more strongly negative than the first.

The emotional salience of a message must also be considered in its environmental context, that is, the norms governing the interaction at hand. Whether those norms are derived from habit, custom, or culture, they always exist. For the cultural case, imagine you must reply to a party invitation in the negative. “Oh, bummer, I would really love to go, but I can’t, since I have to leave early tomorrow. Hope you have a great time!” The semantic meaning and social overtones of the response would be well understood by a typical American. That same response might be met with confusion and even mild annoyance by an Estonian, though. The average Estonian response might be, “Can’t make it, sorry. Cheers.”

So the positive or negative intent of a message guarantees little about how it will be received, and the interpretation of a message relies heavily on norms and information present outside the immediate interaction. Therefore the American and the Estonian’s statements employ context-dependent grammar, in that the individual contributions of each word are not what give these statements their overall positive or negative connotations; it is the arrangement of words within the sentence and the existence of that sentence within a broader context of norms.

For digital communications these norms are exactly the engineering constraints of the system; a certain ISO standard might describe the how to delineate the beginning and end of a message, and both sender and receiver need to be aware of the chosen standard in order to parse the response of the other. To get comfortable in the world of communication, then, one requires a great deal of exposure to new sets of norms.

The flexibility to parse and respond appropriately to any kind of message relies on having a diverse enough set of backing “libraries,” enough training data, enough model input — and many, many attempts at perfection. It may never come, but by definition much will be learned along the way.