Friday, July 16, 2021

Things They Should Invent: concordance tool with a Boolean NOT function

Many words, terms and phrases have a common go-to translation, but also have scope of meaning that doesn't fall under the common go-to translation.
 
If the common go-to translation is extremely common, it can saturate concordance tool results, completely burying alternative translations. This can lead inexperienced translators to conclude there is no other possible translation (even if the go-to is inappropriate), and can even stymie experienced translators ("I know there's another translation, but I'm completely blanking on it!")

To remedy this, I want to be able to apply a Boolean NOT function to the target-language results, to eliminate the common go-to translations and see what's left.

Examples:
 
- Show me translations of porte-parole that are not "spokesperson".
- Show me translations of intervenant that are not "intervener" or "responder".
- Show me translations of animateur that are not "animator" or "facilitator".
 
With the common translations that I know are not suitable out of the way, the tool can better do its job of giving me options to pinpoint le mot juste for my particular translation needs.


I have no idea how feasible this would be from a programming perspective. I know a Boolean NOT can be used in user input, I know that you can filter output by selecting and unselecting attribute tags from a given list (like you often find in online shopping), but I have no idea about the feasibility of filtering output with user-provided Boolean operators.

If it would in fact be unfeasible, I have an idea for an alternative: sort results in alphabetical order by how the word/term/phrase in question is translated in the target text.

This would group all the translations I know I don't want to use together, making it easier to find other options.

For example, if all the instances of "spokesperson" are together (with variants like "spokesman" nearby), I can start at the beginning of the alphabetically-sorted results and scroll through until I hit "spokesperson", seeing all the available options. Then, when I hit "spokesperson", I can jump to the last result and scroll through in reverse order until I hit "spokesperson" again, thereby quickly getting an overview of all the non-"spokesperson" results.

Concordance tools do tend to provide a sentence or a snippet as output, but they "know" what the matched term is, so it seems like it should be feasible to sort alphabetically by matched term but still show snippets.

No comments: