Background on Text

Wanna know something cool?

Right now, you — yes, you-the-one-perusing-these-lines-here you — you are creating the lines as you read.

Cool, huh?

It is really true that readers actually make a text textual. That is to say, it is on account of a reader — like yourself, here and now — that any text even becomes what it was destined to be: Words making sense. Because that is the effect which the act of reading has on text — it makes the silent unread words appear as something we can recognizably say is an example of text. And interestingly, this effect of the reading is, really, diametrically opposed to the effect which writing has on text. Because with writing — there a writer is only about the task of making the words make sense. It's really an act at one remove from text.

You don't agree? Well then, let me ask you this. When you write a few lines and want to know whether you've got there what you mean to say, what do you do? You read the lines, right? There is my proof. Essentially, the features which a writer is actually engineering into a text, features such as coherence and sense — these are precisely the features which the user of text, the reader, does or does not experience. Therefore, even a writer must become reader if he or she is going to know the sense made by the words on a line. And so I say, the real creator of text is the reader.

That's just a cool thing, isn't it?

Now for a helpful thing.

Think back. Do you remember how the system Given-New assigns weights to the Information of every clause? Well, those weights are, when viewed through a reader's eyes, just the answers to questions. Accordingly, the heavier a weight, the more definite and satisfactory the answer provided by that Information unit.

Think of it like this. If the reader is there asking questions of the text, well then, the heaviest-weighted unit in each clause will be — to the reader's mind, anyway — the fullest answer the text has then on offer. That is how the reader-creator makes sense in the act of reading, because this creator-reader asks of the text questions. Readers want to know, and the answering provided by the text unfolding is one particular driver of this textual creation I am talking about.

Take, for example, yourself, here and now.

You want to know what I mean when I say, "Readers make sense of text," and in actual fact, that very effort of yours is the true driver behind this text here. Your will to understand is what's making this blog post appear to you as a whole and coherent text.

Now, of course, I am anticipating your every thought on the topic — I mean, that's my intent, because I am the one who has put forward the topic of sense in text. Nonetheless, you are the one asking the questions and you are the one finding whichever answers are on offer here in this post. Basically, your inquisitiveness is the energy I've tapped in order to write this post. And this is the way it goes with all published texts.

Let me show you what I mean.

We have that much-parsed paragraph from Strong and Efficient Cache Side-Channel Protection Using Hardware Transactional Memory (2017 USENIX). Well, let's put that paragraph back in the paper — let's give the paragraph again its co-textual surroundings, and let's see there how the paragraph and all the lines above and below the paragraph offer the best possible answers for a reader at that stage in the text. To this end, I have parsed the entire Background section, highlighting in red the Focus, that is, the informational weight W3. That is the answer which the reader most wants, in that exact moment. Here is the parse.

Do you see what I am seeing? In the unhighlighted stretches, the words begin raising questions in readers' minds, until in the red, the reader finds the answer. For example,

We now provide background... What on? —— on cache side-channel attacks and hardware transactional memory.

Modern CPUs have hierarchy of caches that store and efficiently retrieve frequently used instructions and data... Sooo... why does that matter? —— thereby, often avoiding the latency of main memory accesses.

The first-level cache is usually the smallest and fastest cache... Like how small and how fast? —— limited to several KB.

A line of text raises a question, then begins answering it, and lastly closes out by giving the fullest answer possible thus far in the discourse. Or to re-express that from the reader's point of view: I begin a clause appreciating the matter at hand, then I become intrigued by the thought arising from the matter, and finally I am satisfied by understanding how this new detail relates back to and connects with how the matter appeared at the start. The pattern is Topic - Question - Answer; the pattern-maker is the reader.

So you see, it lies within the interpreter's judgment to understand a point or no. Textual coherence is not an objective phenomenon to be discovered out there in the external world of print, but instead it is an act of re-creation at every new reading of a text. This explains why a writer can tinker endlessly with a text. It's because each next reading, even of the same unchanged lines, makes new sense of the same words. Likewise, the re-creative energy of the reading also explains why readers of the same text will disagree over what is said there — a perennial source of frustration for authors, but clearly, not a problem they always have the power to fix. Authors have a lot to say, but of course, they have to choose, and when they've chosen this one detail and chosen to reveal it at this exact moment, that may just be the wrong choice for some or even many readers. Discrepancies like these between readers are legitimate. So, the thing to note here is how these discrepancies create different texts out of the same paper. One set of readers will see a paper that is accurate and convincing, and another set of readers will see a paper not to be worth their attention. That is the power which reading wields over the communication of science.

Please email comments or questions to daniel.shea∂kit.edu

This blog is for you.