My grandmother would be about 125 years old today. She was born in a country that didn’t exist when she was born. Now it is called Belarus. English was her third or fourth language. I don’t know if she spoke any of them well, but she got her message across.
She died before smartphones were invented, yet she taught me to read the phone messages I get that are converted into text messages.
Back in the day, she used to write me letters. I recall getting one shortly after I was married. I passed the letter to my wife who exclaimed “What language is this? It isn’t English!”
“Yes it is,” I replied, “but you have to read it out loud.”
When my grandmother wrote “ve vas dere fer dot” – if you read it out loud you understand “We were there for that” – Simple. Her spelling was phonetic, but with a thick Eastern European accent, her accent. “Vent” is the past tense of “go” and so forth.
So when I got a converted phone message from a friend saying “I told him you’re welcome to visit my blog Catherine Crow.” I knew he was telling me about his blog Cantorbury Tales. Not that he was speaking with someone called Catherine Crow. But the misunderstanding and approximate fit is typical for these kinds of transcriptions. It happens all the time. AI seems to always try to correct something but often misunderstands.
It is a reminder that pronunciation, subtlety, and unfamiliar words are not yet incorporated into these transcriptions. If Shakespeare had AI, he would have written “A plaque on both their houses!” and Romeo and Juliet would have ended with brass plaques being installed on the Montegues’ and Capulets’ homes.
The easy way to understand these transcriptions is just to read them out loud and imagine someone with an accent is doing the speaking.
10 Q (which for those who don’t read this out loud, it says “tenk you”)