Open Source, too!
Carol Chapelle at Iowa State has morphed her focus from tech to AI and language learning. A logical change. She is leading up a team with a new journal, Exploring AI in Applied Linguistics.
https://www.iastatedigitalpress.com/plugins/books/154/
Nowhere better can you find the nuances of culture than in the words that are used. Nancy Friedman tracks those at Fritinancy. This year we get a bonus, wordS, not just word.
My favorite: The Fall of the Broman Empire, on the X-odus of people from Musk’s vanity website. Others include 4B (from Korea), Debank, social tonic (laced with THC), Pretendians. I have a couple of young HENRYs in my family tree. And there is the Hawk Tuah girl.
I can tell Nancy has been around a while at this game because she is writing on TypePad, a platform that had its heyday many years ago.
As a slow learner of written Japanese I am at Wanikani Level 20/60 after about a year. I like to check out how others are going about it. So I create a feed for Reddit’s LearnJapanese. Here is someone who finished up all 60 levels of Wanikani (a kanji learning program, $100 off this week for a lifetime license) after only a year and a half, and has now gone beyond Wanikani for a year.
Another interesting student is a German software developer and how he approaches the language.
From that we can learn about new tools. An old one that has been updated recently is Yomitan, whose browser extension allows you to research meaning, readings, and more while on the web. Another new one for me is the jpdb, a database of kanji you can customize to a great extent.
People are talking about podcasts created by NotebookLM. We are astounded at how the two people discuss any topic we want, in such a natural dialog. The male voice sounds like a mild version of Howard Stern, and the female voice reminds me of…not sure. On the other hand, we complain that the voices are always the same, the speed and delivery also the same.
Google has another tool that lets you have more control over the output. Marc Watkins explains.
Right now you can write a prompt or link to a webpage for input, and select voices and dialog register (formality). Still no speed settings.
