ChatGPT for Technical Writing: Simplifying Terminology Investigation
ChatGPT can significantly reduce the time required for terminology research.
The core concept of a useful technical text is the terminology. It is the main content scaffold providing a communication frame and means to exchange information with your readers. Without consistent and clear terminology, we create chaos requiring rewriting by the next release. Latest.
However, finding the right term might be challenging especially when introducing a new topic, shifting to new technology or providing a more detailed level of description.
ChatGPT is a great tool for extensive terminology research. Finding the right term (e.g. differentiating between proxy and gateway) usually depends on the context that is defined, no surprise, in other docs.
At this point, the in-depth research becomes like learning a new word in a foreign language. We are looking for the spot that matches our own use-case following the motto They use it in this way, maybe we can use it in the same way… To know how to use a term, we must research a lot to be able to evaluate the context and see the similarities between our and the reference content.
The problem of analogy
However, there are two weak points of this approach:
- It might be not obvious at first glance, what is the right reference documentation
- The evaluation that the similarities are sufficient and the differences just marginally requires a lot of context and technical knowledge
ChatGPT can overcome the first problem while processing almost all the written content. Providing outputs based on statistically relevant patterns in a text corpus, ChatGPT acts as an extremely powerful corpus manager. It only depends on the right prompt to be able to get a useful answer. For instance, the following prompts work for me:
- Act as a Technical Writer for the web-based graphical user interface. I need to describe a list box offering me some values. Give me list-box-related terminology (types etc.).
- Act as a software expert. Is the gateway the same as the proxy? Tell me in which way they differ and in which way they do the same stuff.
Reading answers to these prompts not as user interface descriptions or proxy definitions but rather as a report about word occurrences mapping the context and revealing relations among semantic patterns is the key.
The answer provided by ChatGPT is not merely a similar case, but rather a synthesized compilation of similar instances significantly reducing the required time for terminology investigations — if you read it as a comprehensive message about terminology relations occurring in the source text.
Just the beginning
However, the critical task of selecting the most appropriate answer still remains with us. It is crucial for every writer to have a deep understanding of their content. Its intended usage usually lies outside the documentation, because it is determined by the target audience, product features or its usage that might be not included in any corpus. ChatGPT does not know anything about our product, use case, context, or intended readers.
However, providing a compilation or extensive synthesis simplifies the evaluation process because we can compare many options and select the most suitable one. What is suitable depends on other criteria than the corpus itself. The turbocharged corpus manager cannot understand the specific intentions behind your writing. Only we can.