Here is another guest post from Bryce Gessell, a scholar who has embraced a Writing is Thinking approach to the writing life. Bryce encountered the work of Klinkenborg and Moi and began to pay close attention to sentences. He helps us write clearer, punchier sentences with his regular contributions to the sentence makeover series.
In today’s makeover, Bryce turns a single 68-word sentence into four sentences. He shows us how to map the relationship between parts of a long sentence by identifying subordinating conjunctions and relative pronouns.
- Thus, in our example, the testing data will be the set of brain activity from the ROI that was not used to train the classifier, and the test is successful if, when fed to the algorithm, the classifier predicts whether a brain pattern corresponds to a stimulus with either a “living” or a “non-living” label beyond some determined decision boundary—which could be chance or certain percent accuracy.
In my last sentence makeover, I wrote about a “typical” academic sentence. These sentences use specialized vocabulary without too much jargon, and are long but not overly so. I don’t think my selection this month is typical in that sense. Instead, the problem with this one is that it’s just too long. The sentence has 68 words, and in the PDF manuscript, those words occupy four complete lines. That’s a lot! Sentences that long occur in spoken conversation all the time, but in written form they are rarer. Good thing, too—long sentences tax readers and are easy to get lost in.
So the fix here is just to make it shorter.
One of the challenges in shortening sentences is unraveling their logical structure. Different parts of sentences depend on other parts in different ways, and cutting a sentence down may require more than just adding periods. Although we might have a hard time discerning that logical structure, it will guide us in making changes—it will tell us what needs to come first and why, what needs to follow, and what we can cut.
We’re lucky that the logical structure of this sentence isn’t that complex. In fact, it won’t take much to decompose it:
- Thus, in our example, the testing data will be the set of brain activity from the ROI that was not used to train the classifier
- and the test is successful if, when fed to the algorithm, the classifier predicts whether a brain pattern corresponds to a stimulus with either a ‘living’ or a ‘non-living’ label beyond some determined decision boundary—
- which could be chance or certain percent accuracy.
These three bullets show the large-scale structure of the sentence. We identify dependencies among different large-scale parts by looking for things like subordinating conjunctions and relative pronouns. Those words tell us that one part of the sentence depends on another for a complete meaning.
In this example, we find five such terms. I have underlined them above—a relative pronoun “that,” the subordinating conjunctions “if,” “when,” and “whether,” and another relative pronoun, “which.” We can zoom our analysis in a bit by marking off these dependencies even further, like this:
- Thus, in our example, the testing data will be the set of brain activity from the ROI
- that was not used to train the classifier
- and the test is successful
- if, when fed to the algorithm, the classifier predicts whether a brain pattern corresponds to a stimulus with either a ‘living’ or a ‘non-living’ label beyond some determined decision boundary—
- which could be chance or certain percent accuracy.
The second-level bullets in this list mark dependent clauses. The relative pronoun “which” also introduces another dependent clause in that third bullet, but at that point it will be natural to introduce another sentence anyway, so I’ve left it at the main level.
We can turn the first independent clause into its own sentence, but as it is now, we have a problem. We need to know what the antecedent of the relative pronoun “that” is—is it the ROI, the brain activity, or the set? The antecedent has to be “set,” though to be fair, seeing that might be tough without some specialized knowledge. We will want to be clearer about that relationship in our makeover.
At 68 words long, the fix with this sentence is to shorten it. The first step in shortening a sentence is to understand its structure and dependencies. We’ve done that with our bullet points above. The next step is to make each large-scale part a sentence of its own. All that requires is adding periods and making adjustments to the beginning and end of each sentence so that each makes a smooth transition to the next.
Here’s what I’ve come up with:
- The classifier didn’t train on one of the sets of brain activity from the ROI. That set will be the testing data. The test succeeds when the classifier predicts whether a brain pattern corresponds to a stimulus with either a “living” or “non-living” label, beyond some boundary. That boundary could be chance or some other accuracy.
From our original single sentence of 68 words, we have come up with four sentences totaling 56 words. That’s a word count reduction of nearly20%, with a big increase in clarity.
To achieve the lower word count, we dropped some of the unnecessary language and re-phrased a few things. The gain in clarity comes from displaying the structure of the idea in multiple sentences. The idea is complex, and describing it in a single sentence would require complexity too. But we can break off part of the idea and present it in its own sentence, making the idea as a whole much easier to grasp.
Each sentence we write should have a single, well-defined job. When one sentence finishes its job, the next sentence picks up our line of reasoning and advances it a bit further. Then the next one takes a turn, and the next. Sentences are much easier to comprehend when they are shorter. Readers understand what one sentence does and why, and then the next sentence takes them a little bit further. Trying to cram everything behind a single period assigns more work to a sentence than it can handle while still being clear. Breaking ideas apart into logical units gives our writing a snappy, clean feel, and helps readers better understand our meaning.
Imagine it this way:
verlyn klinkenborg, Several short sentences about writing, pg 3
One by one, each sentence takes the stage.
It says the very thing it comes into existence to say.
Then it leaves the stage.
It doesn’t help the next one up or the previous one down.
It doesn’t wave to its friends in the audience
Or pause to be acknowledged or applauded.
It doesn’t talk about what it’s saying.
It simply says its piece and leaves the stage.