Sourcing Expressions
Phrasing is capable of turning any text into Expressions, and teaching those Expressions in a manner you'll never forget. The question then becomes where do my Expressions come from?
That answer is going to be different for everyone, but luckily Phrasing makes it easy to import Expressions from a variety of places. On this page, we'll cover:
Importing Expressions from digital text (ebooks, pdfs, articles, webpages)
Importing Expressions from audio (movies, tv shows, podcasts, conversations)
Importing Expressions from music
Importing Expressions from images (physical books, screenshots, signage)
Importing Expressions from LLM conversations
Importing Expressions from failed attempts
But first, an editorial: where do my own Expressions come from?
Expressions can come from anywhere. In an ideal world, every Expression you learn is something you will use often, and exists in some material you could consume over and over (music, tv episode, book) to help you acquire the Expressions after learning it.
Every person is going to have different habits, hobbies, and interests, but my own personal order of preference is as follows:
Something I tried to say, but was unable to correctly/completely: These are my favorite Expressions. Things that I've already felt the need and lack of knowing. I'll often work with LLMs to get the Expressions just right (often trying to condense the pieces I didn't know as densely as possible). These account for 5% of my Expressions.
Quotes from binge-able native material: Expressions that come directly from a book, movie, tv show, music or any other content I would concievably "re-watch" make up a majority of my studies. These Expressions have substantial payoff, because they will inherently dove-tail into the acquisition side of learning. These account for 60% of my Expressions.
Native verified tidbits: These might be expressions I find online, famous quotes, or little bits I happen to overhear. These can be interesting to learn, but they lack the directionality and purpose. These account for 10% of my Expressions
AI generated Expressions: These are just pure generations, either from prompting Sterling or imported from any other LLM. Sometimes, these are generated around topics or role-playing, but many of them are arbitrary uses of some word, phrase, or structure that I'm struggling with. LLM generated Expressions are really fine to learn by, but they can be a little soulless, and they are the least 'sticky'. Therefore they're best used when they are hyper-targeted to a single purpose. These account for roughly 25% of my Expressions.
Importing Expressions from digital text (ebooks, pdfs, articles, webpages)
Importing Expressions from text is incredible easy. Simply copy and paste into Sterling, then select Import Expressions. Fill in the rest of the form, and click submit.
Importing Expressions from audio (movies, tv shows, podcasts, conversations)
Importing Expressions from audio is now supported natively in Phrasing under Sterling → Attach Files → Record Audio. Click the Record Audio button once to start recording, then click it again to stop recording. For the best performance, start recording slightly before the portion you'd like to transcribe. Maximum recording length at the time of writing is roughly 60 seconds.
This feature is intended to be used in a targeted manner. For longer form audio, please stay tuned for a future update.
Importing Expressions from music
Whenever possible, I prefer to find the lyrics online and import it as digital text. If I can not find the lyrics online, then I'll import the audio using the method above.
Importing Expressions from images (physical books, screenshots, signage)
Importing Expressions from images is now supported natively in Phrasing under Sterling → Attach Files → Upload File. This will ask you to upload a file (can be an image or audio file) or take a photo with your camera if you are on mobile.
Most mobile phones will also automatically integrate with OCR in their camera. Both iPhones and Androids offer a "copy text" button in their camera, which is substantially faster than taking, uploading, and analyzing images.
IMPORTANT: Please double and triple check any text to speech before extracting!!
Importing Expressions from LLM conversations
Many of the largest model providers (Mistral, Claude, Grok, ChatGPT) are happy to burn billions of dollars to give you the best quality models for free. For that reason, I'll often use external LLMs to come up with a list of sentences I'd like to learn. I then ask it to print out the Expressions with a prompt like the following, and copy and paste that to Sterling → Import Expressions.
