This post describes my second attempt to leverage ChatGPT to write code that I can adapt for Spanish vocabulary games. My coding skills are next-to-nil, however I am developing a rudimentary ability to see the structure and swap out parts. For instance, the I changed the stock color of the title and "Submit" button, for sleeker ones. More significantly, I created a framework to add words and to possibly give players a choice of what category of words they want to target. I am not going to put Duolingo out of business any time soon. That is not the point, of course. The point it to use available tools to approach language learning in innovative ways. Additionally, the point is to leverage algorithmic thinking to open new avenues for language learning. Ultimately, computer code and human expression are languages with syntax, style and expression in common. Much as we want to divide the studies, they grow from the same roots and support the same branches.
All is not lost and even Bill Gates came out recently saying that he is not overly worried about the risks of AI. I see huge upside for the protection and promotion of world languages. So what does that mean? I am going to keep pushing, keep asking questions and keep looking for ways to AI to augment traditional methods.
Following up on previous posts on HTML language games and inspired by my reading of the new novel “The Centre”, I set out to create a sort of global guessing game. The set up is simple enough: identify words from a list of three possible languages. Is geweest Spanish, Dutch or Bulgarian? What about padre? Ironically, creating the game turned out to be easier than posting a high score. I guess my Instructional Technologist self is working this summer in ways that my language teacher self is not. Maybe I’m just a poopy polyglot.
To create the shell of the game, I repurposed some HTML that I created with the help of ChatGPT for Spanish 1 review game. I swapped out those vocabulary words for ones that I had created in one of many word maps. See tennis, rain and library as samples. I then reformatted those inside Google Sheets with a =CHAR(34)&A2&CHAR(34) and a =CONCATENATE(D2,”: “,E2,”,”). The first put quotation marks around the language and the word an the the second created a line I could use in HTML. At this point I had my first sociolinguistic decision to make. Since the map data renders a language like English or Spanish spoken in more countries more times, I needed to decide whether I wanted those to come up more often in gameplay or not. I landed on “no,” opting for just one word per language. I achieved that with a =countif(A:A,A1)>1 statement in Sheets and then some editing.
So here you go, world, a game of world languages!
As I said, even though the words come from a finite set, ones that I created, I have not posted a perfect score. I’ll keep playing.
In terms of the code, I’m at my top end with the prefab HTML and a few tweaks in terms of feedback, color palate and progress bar. Where would I take this if I could? I’d like to tweak the code to display the right answer if possible. I would add a leaderboard for all players, maybe with some sort of badges. I would add a dashboard for individual players. I would keep metrics on how well–or not–players did with certain languages, looking at this on the micro (player) and macro (all players) scale. I would keep metrics on which languages players confused most often. I would like to add a check so that identical words in different languages don’t cause confusion; for example set it up so esfera comes up with either Spanish or Portuguese as an option, but not both. I would like a better randomizer, because it seems like the same words are coming up. I would additionally like to automate the process of including and translating words. Right now we are at 763. I’d certainly like to scale that. [UPDATE: I made this happen with a list of almost 10,000 stopwords in 26 languages. See v.4.0.] I would like to find a more elegant way to reference data sets and/or arrays of words in different languages. I have also through of flipping the game, that is, showing the language and having players choose from a list of possible words. This last one would be simple enough, I think. [UPDATE: I made this happen and it worked as expected. See v.3.0.] The highest high end upgrade would be creating this game with audio samples. There are certainly sets available, but I don’t have the tools to move past text at present.
Three last non-technical question that come to mind: Is reading and recognizing, reading and repeating, a word a step towards respect? Is looking at language like this furthering learning or feteshizing a language? So much cash and computational power has gone into training LLMs recently… how can we use these technologies to help humans grow their language skills?