This blog is about language translation.
For more than 20 years I was a partner in a Japanese social science think tank. I won't describe it here, that will be a separate blog, because we don't have any comparable think tanks in the United States.
What I wish to comment on is how my partner, Seiji Takekawa, and I communicated. This is highly relevant to machine translation.
If you know any languages or regularly see translations from other languages on Google or other computer programs, you will note that some languages translate fairly well and some languages do not translate at all.
Japanese to English is a complete failure. Machine language translation for many languages is a farce. I do not see any possible improvement on the horizon.
That is why I am describing how Takekawa san and I communicated for two decades.
Early on Takekawa san brought in a translator. His English was poor and my Japanese is equally poor.
After talking with each other for about 5 minutes the translator would be dismissed. We could communicate perfectly.
The way we did it was for him to make a statement or describe a business project in broken English and for me, in poor quality Japanese, to repeat back what I heard or understood.
Sometimes he would talk in Japanese and I would repeat back in English.
Invariably we understood each other and carried on conversations for hours. We carried on conversations that dealt with complex social science issues.
It was very rare for one of us to repeat what we had said in a modified form to make sure we were understood. Most of the time our interpersonal communications were perfect.
To give you an example of a particular subject matter and the level of complexity we dealt with. One of our clients was a religious movement that was interested in selling a new design of household ‘memorial cabinets’. In Japan most households have a cabinet with photographs, incense and fruit dedicated to the memory of deceased family members. Our business challenge was to understand this new religious movement and translate the issues into design criteria for the memorial cabinets and invent new sales channels that could be used. Not a simple subject. Not a simple language issue. But we did it. Our clients were very happy.
We had no trouble understanding each other. I had to describe burial and mortuary procedures in the United States. Not easy for someone, like Takekawa san, who had rarely been to the U.S. to understand.
What this tells me is that complex language translations are quite natural for human beings but not susceptible to computer brute force or the prominent fantasies of computer artificial intelligence.