Machine Translation

For this experiment with AI translation, I decided to use two famous haiku: "The Old Pond" by Matsuo Bashō and "Everything I Touch" by Kobayashi Issa. To maintain as much consistency as possible, I used Google Translate by inserting the poem and copying the result directly and asked ChatGPT to "please translate this to english: [haiku]". The results are as follows:

"The Old Pond" original poem in Japanese: 

古池や

蛙飛び込む

水の音

ChatGPT translated it as:

The old pond—

a frog jumps in,

the sound of water.

Google Translate translated it as:

An old pond; a frog jumps in; the sound of water.

I find it very interesting that both machines translated the poem almost exactly (aside from formatting) despite a lack of depth in meaning. The final line "the sound of water" sticks out to me because although it is a literal translation of the Japanese, the nuance of the poem refers to a particular sound - perhaps a splash. Evidently, the machine did not pick up on that underlying meaning.

"Everything I Touch" original poem in Japanese: 

古郷や 

よるもさはるも 

茨の花

ChatGPT translated it as:

My old home—at night and by day, the wild rose is in bloom.

Google Translate translated it as:

My hometown, night and spring, is filled with thorn blossoms.

These results seem to demonstrate the machines confusion in other ways. Each phrase is translated differently and again does not carry the nuance meant to evoke emotion in the reader. Google Translate's translation "night and spring" is interesting to me as well - perhaps it focused on "はる" and not the overall phrase?

Tori