It took a little longer than expected, but I finished up the paid Substack grammar posts for February 9 and 13. All I have left now are February 16, 20, 23, and 27. In theory, and with a bit of luck, I ought to be done with all of those by this coming Tuesday. The last unit I did today was on conjunctive adverbs (however, moreover, furthermore, thus, therefore, etc.), and that's the final unit of the Adverbs part of this curriculum. The next three units are all about Conjunctions—coordinating, subordinating, correlative—and the unit I'll be ending on is the first unit on Interjections. When I was researching interjections, I didn't know they subdivided into primary/secondary, volitive (expressing a will or wish), and emotive. That's going to be interesting to write about and create examples for. But to get through February, I need only do the first category of interjections: primary/secondary. These are actually two distinct categories, but I'm lumping them together on my curriculum calendar. Primary interjections, it turns out, are pure interjections that are onomatopoetic in nature: Ow! Psst! Hmmm... Secondary interjections, by contrast, come from other parts of speech but are used as interjections: Damn! Fuck! Shoot. Darn. Well, decidedly! And so on.
It occurs to me that I'm way behind in generating quizzes and tests for my Substack curriculum. I've got quizzes up to Concrete and Abstract Nouns, but I've created 45 units since then, so in theory, that means I need to generate 45 quizzes (if I can do six per week, I can be mostly caught up in seven or eight weeks). I have eight sections (several units per section) done thus far, so on top of another 45 quizzes, I need to make eight large tests. The quizzes, as you know, are five questions each and are purely multiple choice, but in the spirit of choose all that apply (or none). The tests, which will be fifteen to twenty questions long, will include matching, fill-in-the-blank, and possibly even sequence questions. I hesitate to do short-answer questions (a much better way to test knowledge than multiple choice) because I need to figure out how to get AI to grade the quality of the short answers, and frankly, I don't think short-answer questions are possible. We've had AI-scored essay tests for years, but because AI isn't actually intelligent, it scores by looking at things like sentence complexity and overall organization (akin to a "reading index"). It's conceivable, then, for AI to give a high score to a well-written irrelevancy. It has little way of knowing whether a student has in fact answered a question... although I admit that that weakness may have disappeared over the last few years. All the same, I'm still hesitant to grant that AI has gotten that good with social awareness and the assessment of accurate context, focused relevancy, topical significance, etc. Maybe it has. Maybe I should test Grok and ChatGPT.
As always—lots of work to be done.
In other news: my subscriber count (free + paying subscribers) has reached the emotionally significant 20 mark (but only thanks to help from a friend who pities me—ha ha). Next step: 50 subscribers. Then 100. Then 500. Then 1,000. A man can dream. With my luck, I'll be stuck at 20 until I kick off from a second heart attack or stroke. We'll see.
ADDENDUM: Here's AI's own answer as to how AI assesses essay quality:
AI scores essays using Natural Language Processing (NLP) to analyze grammar, structure, coherence, vocabulary, and argument, learning patterns from vast datasets to provide rapid feedback, often achieving accuracy comparable to human graders for low-stakes tasks but still requiring teacher oversight for nuanced, high-stakes evaluations. These systems assess logical flow, supporting evidence, and stylistic elements, flagging errors and suggesting improvements instantly, though they might miss deeper meaning or context.
In other words, it's still not a human-level assessment of someone's writing, so I'm not sure I want to include AI-scored, short-answer questions in anything I create for Substack.





No comments:
Post a Comment
READ THIS BEFORE COMMENTING!
All comments are subject to approval before they are published, so they will not appear immediately. Comments should be civil, relevant, and substantive. Anonymous comments are not allowed and will be unceremoniously deleted. For more on my comments policy, please see this entry on my other blog.
AND A NEW RULE (per this post): comments critical of Trump's lying must include criticism of Biden's or Kamala's or some prominent leftie's lying on a one-for-one basis! Failure to be balanced means your comment will not be published.