Advanced Business Writing
Clarity and Flow

Good Modifiers

In review, sentences that flow well are often written in the format:

S~V~O~i

S = Subject
V = Verb
O = Object
i = information

The goal suggested in a prior lesson was to get the reader quickly from the subject to the verb and then to the object, with extra information following all of that. Long sections of information at the end of a sentence can make that sentence both informative and elegant. However, first drafts sometimes don't flow well because they are written in such a way that readers have to temporarily store information in short term memory as they strive to understand the whole sentence:

Of all the disciplines of artificial intelligence that are important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, which requires continually learning new tasks and remembering how to perform all of them in a future that isn't plagued by catastrophic forgetting, which is a problem because they are incapable of sequential task learning, which is due to the fact that the weights in a network that are needed to solve for task A are changed when the network is then trained to solve for task B.

Short-term memory is taxed when there are too many clauses starting with the words that or which:

Of all the disciplines of artificial intelligence that are important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, which requires continually learning new tasks and remembering how to perform all of them in a future that isn't plagued by catastrophic forgetting, which is a problem because they are incapable of sequential task learning, which is due to the fact that the weights in a network that are needed to solve for task A are changed when the network is then trained to solve for task B.

If you find more than four uses of the words that and which, you have good reason to edit for better flow.

There are several ways to do this: absorb, elaborate, or detach.

Absorb

Bring it in as a modifying phrase.

Of all the disciplines of artificial intelligence that are important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, ...
[a bit taxing on the brain]

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks that generalize in real-world environments, ...
[flows better]

Elaborate

Expand on the topic or statement.

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, which requires continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting, ...
[taxing on the brain]

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, a technology made possible by continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting, ...
[flows better]

Detach

Break it out as a separate sentence.

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, a technology made possible by continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting, which is a problem because they are incapable of sequential task learning, ...
[taxing on the brain]

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, a technology made possible by continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting. Such forgetting remains a problem because these networks are incapable of sequential task learning, ...
[flows better]

Elaborate (once again)

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, a technology made possible by continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting. Such forgetting remains a problem because these networks are incapable of sequential task learning, which is due to the fact that the weights in a network that are needed to solve for task A are changed when the network is then trained to solve for task B.
[taxing on the brain]

Of all the disciplines of artificial intelligence important to natural language understanding, few are more promising than networks with memory that generalize in real-world environments, a technology made possible by continually learning new tasks and remembering how to perform them in a future that isn't plagued by catastrophic forgetting. Such forgetting remains a problem because these networks are incapable of sequential task learning because the weights in a network that are needed to solve for task A are changed when the network is then trained to solve for task B.
[flows better]

Lesson: Good Modifiers
Module: Clarity and Flow
Course: Advanced Business Writing