2 Questions I've been asking myself all these years

 Tomorrow is the last session of the MA course on 'Professional Development in English Language Teaching." I will end the course by sharing my own answers to the following 2 questions that I've been asking myself all these years. 

Q1: Would I still be so enthusiastic about teaching today if I had not started out as an ENGLISH teacher? 

--- In other words, what makes English Language teaching so fascinating? 

Q2: What kind of person would I be like today if I had not started out as a teacher? 

--- Because teaching has shaped the person that I am today.

Pure listening, or listening while staring at the screen

 今早的課堂我做了一個實驗， 我把學員要聆聽的audio file拆做兩部分，上半部上傳到Spotify 作為純聆聽的podcast episode (Option 1),下半部我用video editing software,加入audio file,然後插入幾張教室的image, 但內容和audio file的內容無關，最後把完成的video 上傳至YouTube （option 2)。

當學員聽過純audio 的option 1, 和聽過embed 入video 的option 2, 我問學員喜歡哪個聆聽的模式，我原本也猜測有学員喜歡option 2, 但結果比我預料的更一面倒：83%學員喜歡option 2。

我不是第一個有此觀察的人，至於為甚麼今天的人，就算畫面和內容無直接関係，也喜歡盯著畫面去聆聽多過純聆聽，這究竟是social habit （YouTube, Facebook Shorts)，抑或有心理學原因，或甚至是neuroscience, 我便不得而知了。

AI apps for creating resources

 早在兩個半小時的課，已經用了4件AI-generated的教材。後來我告訴學員哪些教材是來自AI, 他們也很詫異。

我說笑我快將失業了。但其實省下製作教材的時間，教師有更多時間設計interactive的課堂和高階思維活動。到最後，關鍵還是我們怎樣運用科技。

https://www.questionwell.org/?fbclid=IwAR39w-WWFCf0LsgN85CBAi9q372oGsvEExq3oywDqgotlKJDtLHZAHs0hSePQ

AI-generated slide decks: When are they useful?

 

in a previous post, I said that all the AI-tools for generating slide decks that I have tried are not satisfactory. This is because:

(a) these apps will not give me a slide deck which fully captures my line of thinking; in a presentation there are usually certain points that I'm driving at, and

(b) I like to use my own examples for illustration.

But I've found that if I simply need a mini presentation for introducing an idea, then such apps can be useful. The other day, I was planning for a lecture on using case studies for teacher professional development. I wanted to start the lecture by mentioning the 3 disciplines that use case studies as an established teaching method and for professional development. Then I wanted to cite one example: The Harvard Business School's use of the case method. I just needed a mini introduction to that.

I went to Canva, a wellknown app for creating visual materials, and used their new AI 'Magic Write' function to get a short presentation on the Harvard example. In less than 30 seconds, they gave me the presentation below, which exactly answered my need.

If I had to create this presentation from scratch myself, searching for the information, distilling the information into main points, creating a presentation with the main points, dealing with the slide deck design, how much time would that take me!?

My tentative conclusion: AI-tools will not give me an immediately usable custom-designed presentation for a whole lecture/lesson, but they are useful for creating short presentations for introducing an idea or topic.

(BTW, I'll be using the above slide deck tomorrow morning. After the short presentation, I will ask the teachers whether they have suspected something about the slide deck.)

The question set export function in QuestionWell

This morning I explained how I use QuestionWell, an AI app, to generate a set of MC comprehension questions from a text. The question set can then be exported as DOCX, or as a Microsoft Excel file. The app allows me to generate two versions: Version 1 with the questions only, and Version 2 with the Answer Key as well.

At this point, I can share Version 1 with my teacher students. They will then read the text and then mark the correct answers on their copied DOCX document. But this afternoon, I decided to go one step further, by turning the question set into a Google Form quiz. I converted the DOCX document to a Google Doc. I then opened the Google Doc extension called Automagical Forms, and had it convert all the questions on the Google Doc to a Google Form quiz. Within 5 seconds, I had the Google Form quiz ready to send.

Imagine if I had to type out the Google Form quiz myself!

Of course one thing I don't understand is: Despite all these time-saving AI apps, why am I still so busy?

(BTW, QuestionWell will also allow you to export the question set as an Excel file, which you can then import into Kahoot or Quizizz immediately to create a quiz game.)

Using QuestionWell to help students focus on their reading

 Yesterday, I said that QuestionWell was by far the best AI tool for generating reading comprehension questions from a reading text. The only concern was that it could only generate MC questions which asked for specific information in the text (in other words: no HOT questions). In other words, don't expect the app will give you all the questions that you need for a full-fledged reading comprehension lesson. (https://www.questionwell.org/

Nevertheless, QuestionWell is still useful for generating a short MC quiz to accompany a short reading activity to help students focus. In an upcoming class session, one of the activities is for the teachers in my class to read a case story, and then get into groups to discuss the case. In order for the teachers to benefit from the case discussion, it is important that they read the case story carefully first. I went to QuestionWell and input the case story. The app generated 22 MC questions for me. I chose 8 of them, and would use them this Saturday to prepare the teachers for the case discussion. It would take me a lot of time if I were to come up with the questions myself. 

In other words, do not rely wholly on these AI tools to create teaching materials for us, but if used with discretion, they CAN save us a lot of time. 

QuestionWell for generating MC question from a reading text

 QuestionWell is by far the best AI-powered tool for creating reading comprehension questions from a reading text that I have tested. The pluses: 

- It can generate a long list of MC questions for you to select the ones you want to use. 

- It tells you the key questions among the ones generated. 

- It also generates a list of learning objectives embededed in the generated questions. 

- You can then export the questions as a spreadsheet, which you can then import into Kahoot, Quizizz, or Blooket immediately, or as a docx file. 

Of course, like the other AI tools for generating reading comprehension questions, QuestionWell is only good at creating questions that ask for specific information. 

https://www.questionwell.org/

However, of the 6 AI-powered tools for creating slidedecks I have tried so far, none of them is good enough. So maybe I can conclude now that as of today, AI technology is good at generating Reading Comprehension materials, but not at creating presentation decks. 

https://www.questionwell.org/

No Hawking vs No Hawkering

 "No hawkering", or "No hawking"?

Strictly speaking, they should say 'No hawking'. The word 'hawking' already exists in English: Hawking is what hawkers do, just like busking (not 'buskering') is what buskers do; teaching (not 'teachering') is what teachers do; painting (not 'paintering') is what painters do ...

But on second thought, 'hawkering' is not necessarily wrong. It can refer to people who like to play hawker once in while - as amateur hawkers. Just this morning, I saw an English-speaking girl gleefully buskering in Citygate Town Square under the supervision of her mum. She was doing a Chinese dance; no one put money in her box; she didn't mind, as she was simply having fun buskering, she was not really busking). 

BTW, I googled 'No hawkering'. The expression IS used in Malaysia and Singapore. 

Veed.io for editing video

 Veed.io is a web-based video editing app. It's simple and quite easy to use. You can export the created video as an MP4 file, or just get the Share link. 

https://www.veed.io/

Balabolka for converting texts to audio files

 Just tried a couple of Web apps for converting texts to audios: Murf.ai, and Speechify: 

Plus: 

- the resulting speech sounds quite natural

Minus: 

- once you want to download the audio to your computer, it asks you to upgrade. 

+++++++

I then tried Balabolka, a desktop app: 

Plus: 

- It's open-source; it's free to download the generate audio

Minus: 

- the resulting speech is not natural enough

++++++++++++++++

I finally got back to my Spotify episodes, which are converted from my Wordpress blog via postcaster.spotify.com (formerly Anchor.fm). 

The audios are natural enough, but the steps for downloading an episode as an MP3 file are slightly complicated. 

True/False interactive format in Wordwall

 I had been using Wordwall for a few years, but I had never touched the 'True/False' interactive format because I thought it was just like the traditional True/False quiz format. The other day during a lesson observation, the teacher used a Wordwall interactive format which I found quite exciting. He had several statements moving from one side of the screen to the other continuously, and near the bottom of the screen were the words 'True' and 'False'. As the sentences were moving from right to left continuously, each sentence would appear on the screen for about 3 seconds only, and within 3 seconds, students had to decide on the answer, and then the next sentence would show up. 

After class, I asked the teacher which Wordwall interactive format he had used, and he said it was 'True/False.'

Later, I gave it a try, and indeed, it was not the traditional 'True/False' I had had in mind. 

So, I've created this one for an upcoming class. The task is to decide, for each of 8 teaching issues, whether it is more suited for action research or for a case study. Give it a try if you're interested: https://wordwall.net/resource/55011264

GotFeedback for generating feedback on students' writing

 

GotFeedback is another tool that has jumped on the AI bandwagon. It claims to be able to provide feedback on a piece of (student) writing. At the moment, there are 4 default areas of feedback (e.g. narrative structure), but users can also use their own custom prompt (i.e., tell the AI system what you want it to comment on). If this AI system works well, it will of great help to ESL teachers. 

I tested it with a few essays. It doesn't work like Grammarly, which highlights problems with individual sentences. Instead, it provides general comments with respect to one of the 4 default areas or on users' custom prompt. The generated comments will cite examples from the essay. To me, the comments are still a bit too general and not concrete enough. Sometimes, it gives me the impression that it has jumped to those feedback conclusions. But maybe that's because the essays I tried are not terrible ones. If I get to find a pre-intermediate ESL learner's composition with lots of problems, I will test this AI system again to see how it performs. 

https://feedback.gotlearning.com/

ChatGPT style

 For me, these ChatGPT-generated essays have a certain ChatGPT-style so that once you're accustomed to such essays, it's not that difficult (for a human) to tell whether an essay is ChatGPT-generated or not. (For example, they like to beat about the bush; they like to talk around an issue instead of developing a thesis; they like to finish with a play-safe counterpoint.) But then that will probably be ChatGPT's next mission - to rid itself of its ChatGPT style.

Test-driving SlidesGPT

 Just had a go with SlidesGPT. The result looks more like a slide deck, in that the content of each slide is in point form (unlike tome.app). You can then download it as a PPT or Google Slide deck, and use those slides which are useful. Overall, the resulting slide deck is still talking around the topic, and is lacking in depth or a coherent development. This seems to be a weakness common to all AI tools at the moment. 

AI apps for generating slidedecks

 Just tried another AI app for generating a presentation slidedeck, called tome.app. Again, the result is disappointing. The text content of each slide is apparently replicated from a corresponding summary essay. The visual that accompanies each slide has nothing to do with the text content. 

I think I can now draw this conclusion: As of today, AI apps are still not useful for generating slidedecks from a topic command. (In case you have come across a useful one, do let me know.)

It's imperfections that make us human!

 最近因貪玩喜歡把上課時部分內容用AI app 製成talking head 的video, 我對學員說我邀請了某專家就這課題作分享。但學員很快便看出video中present的不是真人，不是因為這talking head太假，反而是他/她無論聲調和表情太過完美！真人說話是不會這麼完美的。

於是我突然有這個epiphany：這些年我被self-help 書籍洗腦，覺得凡事都要力求完美，We should forever strive to be our best self, 卻原來It's imperfections that make us human!

Trying out Conker.ai's other input mode

 Then I tested Conker.ai with the other input option: I gave it a topic and asked it to generate 5 MC questions for me. The topic was 'action research for teacher professional development'. 

The resulting questions are not bad. Of course, again, they are all about specific information. But I can use the quiz in the lead-in stage of a lecture (e.g., as an entry ticket), in order to motivate interest, or assess students' prior knowledge. 

One plus with this app is that the generated quiz can be assigned to students immediately, or exported to Google Forms. 

Test-driving Conker.ai

 Conker.ai is an AI system for generating quizzes. You can starting by either (a) giving it a topic, or (b) providing it with a reading text. It will then proceed to generating the number of questions (so far, MC questions only) that you have requested for you. 

I tested it with input (b). I provided it with a reading text, and asked it to create 5 MC questions for me. 

The result: 

1. Not bad; the MC distractors are sensible. Hence students do have to read the text to get the right answers. 

2. The questions all involve 'looking for specific information.' None of the questions tap higher-order reading skills (e.g., inferencing). 

Test-driving Curipod's AI lesson generation function

Curipod is like Nearpod, Pear Deck, Quizizz Lesson, and Kahoot Lesson: an app for creating Interactive Lessons. But what distinguishes Curipod from the rest is its Al lesson-generation functionality: Give it a task (e.g., a topic), and it will generate an interaction lesson for you.

I couldn't wait to have a go, asking it to create an interactive lesson that elucidates the difference between WILL-future and GOING TO-future in English. Luckily, the result is disappointing.

The major problems with the AI-generated interaction lesson:

- The content slides are just beating about the bush, rather than addressing the heart of the matter.

- The activity slides rely too much on open-ended drawing, which do not stimulate meaningful thinking or learning.

- It simply doesn't know how to scaffold learning.

Hence, we English teachers won't get to lose our jobs too soon