The purpose of writing is to communicate to the reader. Since both writing and reading are lossy processes, we want to write in a way that:
It turns out that syntactic structure directly affect the reader’s ability to easily and accurately decode what’s on the page!
This post is a simplification of The Science of Scientific Writing, which draws from the fields of rhetoric, linguistics and cognitive psychology to argue that readers interpret writing based on its syntactic structure.
As a trivial example where syntactic structure is important, consider the following information:
t(t)=15, T(temperature)=32;
t=0, T=25;
t=6, T=29; t=3, T=27;
t=12, T=32; t=9, T=31;
This piece of writing conveys temperature over time. However, consider this syntactic reorganization:
time (min) temperature (C)
0 25
3 27
6 29
9 31
12 32
15 32
It’s now clear that time is sampled in 3 minute intervals, and that the temperature is rising linearly over time (The field of data visualization goes further to instead encode this data as a visualization).
Note that you read the data from left to right. This is because you expected a certain structure:
This left to right flow is a common pattern that readers expect when they read english:
Expectation ---> Discovery
or
Old info ---> New Info
Think about when you read a paper. You expect it to follow a certain structure of Introduction, Methods, Conclusion, with Related Works thrown in after the intro or before the conclusions. If the paper is not organized this way, it instills confusion and forces the reader to work harder to figure out why.
Writing is composed of units of discourse. A unit can be a clause, a sentence, a paragraph, a section, or an article. The most important principle is that a unit conveys a single point. Consider the following examples:
Note that an article is composed of sections, which are composed of paragraphs, and so on. Thus, you can think of each unit (say, an article) as a tree of smaller units. Each tree makes a single point, and each node’s point is argued in more detail by its child units.
Consider a unit at the granularity of a sentence. We will discuss how people rely on its syntactic structure to decode a sentence, and what readers expect when reading from one sentence to the next. We use sentence as the example, but the principles apply to units of any granularity.
These principles are:
As a example, in the sentence Spot chased the cat, the subject refers to Spot, the verb is chased, and what we learn is that the subject chased the cat.
Once the reader has found the subject, she literally waits in anticipation for the verb. Without the verb, the reader doesn’t know what the subject is doing. This penalizes the reader in two ways:
Consider this mess:
Once the subject, the part of the sentence that describes what the sentence is about or who performs an action has been found, she literally waits in anticipation for the verb.
The issue is that the subject and verb is separated by over a dozen words; the reader needs to wade through this mess to find the verb. A simple reorganization brings the subject next to the verb, and simplifies the sentence:
Once the subject has been found, she literally waits in anticipation fo the verb. The subject is the part of the sentence…
The reader mentally inhales when she starts a sentence (or any unit), and exhales at the end. When she starts, she expects a reference to something she already knows, or has already learned earlier. When she finishes, she expects to be rewarded with new information.
The end of the sentence is called the Stress Position. The reader expects that the point is at the end of the sentence (new information). In the following example, the reader is likely to think that Spot is a West Nile Lion.
Spot, a big yellow dog with a brown spot, bit the neighbor’s cat that was found to be similar to a species of lion in the West Nile
The reason this happens is because the sentence has multiple points that it is trying to make. Remember: each unit makes a single point! To fix this issue, we can syntactically introduce additional units, one for each point. Note that a semicolon separates two related thoughts, where thought can be considered a standalone sentence:
Spot is a big yellow dog with a brown spot. He bit the neighbor’s cat; the cat was found to be similar to a species of lion in the West Nile.
Finally, the start of the sentence is called the Topic Position. The reader expects something she already knows (old information). Thus, it should either reference 1) a previous topic position/subject or 2) a previous stress position. Note that this is typically where the reader expects the subject of the sentence.
In the following is an example where the topic position of the second sentence refers to the stress position (cat) from the previous sentence. This is an example of flow.
But if the first sentence is different, we lose the flow. The cat does not refer to anything in the previous sentence!
We can fix this by making spot the topic of the second sentence. Now there is a connection (flow) between the two sentences.
This is why passive sentences are discouraged. They often move new information into the topic position. Remember, new information (the surprise) should be in the stress position. For an example of an extreme abuse of the passive voice, see this article.
These principles are very simple, but apply consistently to sentences, paragraphs, and even sections. Keep in mind that these are syntactic expectations, and do not deal with semantics. Adhering to these principles is needed to make the content easier and unambiguous to decode, but what the content is needs to make sense as well.
When writing (or communicating in general) ask yourself these questions:
To summarize, here is a helpful diagram of everything we discussed:
