The challenge of grad school is becoming informed enough about what you are really interested in to be able to choose a program that fits- without, as you noted, getting too narrow too early- especially when you are looking at a boundary-crossing field. I hadn’t realized that started with CS, and are now adding linguistics, so you will need some linguistics classes to help you work out what you find most interesting!
Grad school admissions is different-and in some ways easier than undergrad, because it is more specialized. For a masters program, it is going to be grades (overall and major specific), test scores (where relevant), LoRs and your Statement of Purpose (how t
This time next spring look at grad school at the details of programs / outcomes.
Like choosing ECs in HS that were joyful to you, when you look at the details of masters programs look for the ones that make you think ‘ooohh- that would be interesting!’ Collect those into a pile and after you have a group go back through them looking for commonalities.
Follow what is most interesting to you. It may take you to that ivy-covered Ivy you have been dreaming of! but even then- you want the ivy-covered Ivy that is truest to what you find interesting, and you only figure that out by focusing on what’s actually in the programs. This is much more important for grad school than for undergrad, which is (in the US) designed to be a relatively general program.
Emily Bender, a prof of computational linguistics at UWa, wrote an article about comp ling & grad school- my post got bounced back for putting the link in but if you do a search for the topic you should find it.