Recontextualisation is a process that extracts text, signs or meaning from its original context and places it in another context. This time we are going to ask Gen. AI to turn a song into a haiku.
A haiku is a traditional Japanese form of poetry consisting of three lines. The first line has 5 syllables, the second line has 7 syllables, and the third line has 5 syllables, following a 5-7-5 syllable structure. Haikus often focus on nature, seasons, or moments of personal insight, capturing fleeting moments with simplicity and depth.
I am going to ask ChatGPT to turn 5 Beatles songs into haikus. I will use this prompt five times with a different song title each time: I want you to turn the song “song title” by The Beatles into a haiku.
Now, can you guess the songs?
Fading memories,
Time lost in silent echoes,
Longing for the past.This haiku reflects the song’s exploration of nostalgia, loss, and the desire to return to a simpler time. The imagery captures the emotional weight of the lyrics, focusing on the themes of yearning and reflection.
Dreams of peace unite,
Worlds without borders or strife,
Harmony in hearts.This haiku reflects the song’s vision of a peaceful world, emphasizing themes of unity, love, and the possibility of a better future. The imagery captures the hopeful and idealistic nature of the original lyrics.
Winter fades away,
Bright light warms the world anew,
Hope in every ray.This haiku captures the song’s essence of renewal and optimism, reflecting the transition from darkness to light and the joy of brighter days ahead.
Feel the rhythm rise,
Joyful shouts and dancing hearts,
Let the music play.This haiku captures the energetic and celebratory spirit of the song, emphasizing themes of movement, joy, and the power of music to uplift.
In my experience Gen.AI does not fare well when asked to generate highly creative content from scratch, such as poems or jokes. However, the more precise and specific the prompt, the better the results. More importantly, when recontextualizing content, Gen.AI performs exceptionally well if it has a clear source to draw from –like the self-contained lyrics of a song and abundant online interpretations of their meaning- and a defined format to follow, such as a haiku with its distinct metrical structure and vivid imagery.
I was rather pleased with the haikus above. If you haven’t figured it out yet, the songs are “Yesterday”, “Imagine”, “Here Comes the Sun” and “Twist and Shout” in that order.
Now, how does this translate into the classroom setting? We could ask Gen.AI to turn a few songs our students are familiar with into haikus. Then we could display printed copies of the haikus in the classroom for the students to read and try to guess the songs. It’s up to us whether to provide the song titles beforehand or not and make it a bit more challenging.
We could create a fill-in-the-gaps activity using printed copies of the haikus with some words missing. We would provide the words at the bottom of the page, arranged randomly. Students would then have to place the words where they belong for the haikus to make sense.
Another highly challenging way to exploit this material for higher levels could be to jumble up all the words from each haiku for students to reconstruct, as in:
longing in memories for time echoes the silent fading past lost
Needless to say, we should probably do some preparatory work before these activities by explaining what haikus are, showing some examples and having conversations with our students around those examples.
To cap it off, we could ask our students in class to choose a song they know or like, look at the lyrics (if they are allowed to use mobile devices), and turn it into a haiku using pen and paper with no technology involved. Then, they can display their haikus on the classroom walls for everyone to read.
Of course, other types of formats can be explored too. I just had the song Imagine turned into a limerick and also into a poem as if it had been written by T.S. Eliot. Now, over to you.
AI in ELT 