Can you teach language analysis by playing?

I recently read this interesting article by Tom Bennett on the limitations of learning through play. I immediately (and sycophantically) responded to say that I agreed and thought play wasn't essential in order to teach language analysis. And it isn't. I'd go as far as to say that there is often no fun had by anyone in many of my lessons. Which, if I'm honest, does trouble me. Am I really as mean as Tom? Isn't he Government Overlord of Telling People Off or something? I started thinking about how I could use play. 

Which made me think about YuGiOh and how brilliant lots of the children at my school are at it. YuGiOh is an essentially indecipherable card trading game played at impossible speed involving intricate, multi-faceted analysis of your own cards, and those of your opponent. I've watched it. It's been (very patiently) explained to me. I've got no idea how it works. But the children do, and they use complex analysis skills, the kind I despair of them ever showing in my classroom, effortlessly, automatically. 

Which made me think about Daniel Willingham and the research finding that "poor" readers who knew about baseball understood text about baseball better than "good" readers. Willingham says that people are, generally speaking, pretty good at understanding things they know something about. David Didau has made a similar point here. Analysing isn't all that hard; it's knowing what to think about that's hard. I can analyse a piece of text because I know lots about what makes a piece of writing good, not because I'm great at analysing. I used to be able to analyse the flight of a cricket ball because I knew what was required to execute an elegant, graceful cover drive (so successfully that I regularly produced such masterpieces once, or even twice a season). But, despite these great analysis "skills", I can't analyse Beethoven's symphonies (or, if I'm honest, anything in the Little Mix discography) because I don't know anything about how music is put together or any of the vocabulary to explain it. 

So far, so dull. I then read another brilliant blog by Fiona Ritson (seriously, all her blogs are brilliant - read them here) about preparing students for Lang Paper 1 by using and deconstructing model answers. Then I went swimming. And while I was swimming I thought: I wonder if they could play YuGiOh with model answers? Trade different pieces of text based on the relative merits of their use of language? "My Dickens scores 5 stars for use of complex sentences!" "Beats my Orwell: 2 stars cause he never uses more than 5 words per sentence!" I mean, even writing it out is fun!

Now, as I admitted earlier, I have no idea how YuGiOh works, so my game is basically Top Trumps. I'm going to write (or find) short pieces of text and, initially, rate them on their use of writing techniques (see the picture for an example - I just did this very quickly, so would appreciate some feedback on the categories! I think that it will be better with more specific categories). The children can have a few each and will play, trying to acquire as many cards as possible. At first I think I'll give them the ratings, so they'll only really be analysing the scores. However, they'll need to read and analyse the texts in the event of a tie. Nevertheless, they will be using the language they'll need to know to analyse the texts themselves when we get to that. When they're comfortable with that language, then I think I'll get them to rate (and write) the pieces of text themselves. Not only can they play, but they could also write up some of their classic plays (like those people who kept diaries of their Risk games): My extract from Henry V was able to defeat the extract from Diary of a Wimpy Kid because.... I mean, seriously, who isn't having fun in that lesson?

At this stage it's just an idea, but I'd appreciate any thoughts or ideas from anyone who has done something similar. Does it work?