Defaulting the gender of adjectives to female when translating first person sentences from English to Spanish

Hi. I have this issue when translating messages from English to Spanish, for instance the sentence:

I am tired

Gets translated into

Estoy cansado

Which is fine, but for my specific use case I want the translator to throw the female adjectives when translating first person sentences, in other words, to show “Estoy cansada” instead (“cansada” being the female version of “cansado”. Is there an easy way to perform this?

Thanks in advance.

There currently isn’t a good way to do this.


Thanks for the reply. I’m working on a chatbot, said chatbot uses an English backend and it’s supposed to be a girl, so I want to always transate the responses using libretranslate and always use the female gender when it’s talking on the first person. I’ve been accomplishing something closer on MyMemory (a different translation provider) by adding extra context to the translation. For the OP case

I am tired

I’m adding the sentence "(The first person is female) turning the input text into

(The first person is female)I am tired

Which results in turning the sentence into the female gender “estoy cansada” after removing the translated text between parenthesis. The drawback is that this way doesn’t recognize if the sentence is followed by a adjective “I am working” and “I am tired” would get the same treatment, resulting in redundant work and it’s not a complete solution, sometimes the translator defaults to male gender nevertheless. On top of that it doesn’t work on libretranslate, adding the text between parenthesis has weird interactions such as removing them in the translation messing up or ignoring them completely and still using the male gender for the adjectives.
Still, I’m surprised my hacky solution works for something like 60% or 70% of instances.

After reading the thread you linked I had a similar idea:

Take a Spanish-English dataset
Remove all the instances of first person sentences with adjectives
Add new data to the dataset that has female first person sentences in Spanish translated into English
Generate the model

Would that work for my use case? how many sentences should I add? How big should the dataset be?