I will bear that in mind …
Quite right, I didn’t take context into account. Consider the accusation of ungrammaticality withdrawn!
It’s sā assa, where sā is feminine to agree with taṇhā: lit. “that (she) of him …”
I think it’s best to represent gender based on the meaning of the text rather than the linguistic forms.
As the previous example shows, it’s impossible to represent Pali genders in English. How could you do it: “He had a her-craving for he-food …”?
Any translation drops the genders maybe 95% of the time, because they simply don’t exist in English. This makes precisely zero difference to the meaning of the text, which is why the long-term tendency is for languages to move away from gendered forms. Like so many aspects of heavily declined languages like Pali, the text is simply overdetermined with grammatical information.
When you translate to English, you can, if you want, represent gender in a very few cases—i.e. personal pronouns—and you can make a stylistic choice to either do so or not do so. The choice to use gendered pronouns is no less political and deliberate than the choice to avoid them.
The cases where this comes up most often are such contexts as the above, where it is an abstract discussion equally applicable to everyone, regardless of gender. In modern English, such cases are usually handled in such a way as to avoid gender bias: by varying the pronouns, rephrasing the sentence, or using the singular they or the generic you.
There are cases where there is a gendered assumption, not just in the language, but in the social roles. For example, we have idioms like purisaviriya, “manly energy”, or, say, a simile where someone chops down wood or opens a business or is a warrior. When I judge (subjectively, of course) that it was a gendered role in that time and place, I keep the gendered forms, .
But in the case of general Dharma teachings on conduct, psychology, and so on, we have abundant evidence that such things are treated identically for men and women. In such cases the semantics of the text corpus are clearly gender neutral, and that’s what I’m interested in conveying.