So it would be just from linguistics? Back in DOS days, I used to have a word processor that would transcribe written pages into several ethnic sounding dialects, including valley girl talk. It could fool someone. I imagine it could also be done without a computer program (although I don't know how) if someone was intent on providing false leads. I wonder if grapho analysts could detect this?
I think that's how it's done. I hunted around to find an easy example, but couldn't find one. But I just happen to have a real life example. This sentence below is the text of an actual e-mail I received this afternoon, from one of my brothers-in-law:
"OK thanks, I habe entered your e-mail in me directory."
This particular brother-in-law of mine spent a year studying in the U.S. about 10 years ago, so he understands a lot of English, but is by no means fluent. Notice the "b" in "habe". In Spanish, "b" and "v" sound almost exactly alike and are often used interchangeably. Notice how he says "me directory". He should have said "my directory" of course, but in Spanish, "mi" means "my" in English, but the English "me" sounds like the Spanish "mi".
If I didn't know who this e-mail was from, I would be able to tell it was from someone who speaks Spanish as a first language.
These kinds of errors can give someone away as a non-native English speaker. The way they may omit pronouns, the placement of adjectives, slang or colloquialisms from their native language translated literally into English. You can pretty much tell the difference between a poorly educated English speaker and someone for whom English is not their first language. Even people who are born deaf may write in a stilted, strange-sounding way, because American Sign Language's grammar is different from English.