Linguists occasionally encounter examples parallel to mistranslated signs like the one Mark wrote about. The situation arises when someone with little or no knowledge of the native language, typically an anthropologist, elicits information such as place names and writes down whatever the response is. When a linguist familiar with the language later reviews these records, some place name will prove to be uninterpretable until the linguist realizes that what has been recorded, usually in a garbled form, is the response "I don't know". There are various stories of this type in linguistic folklore, and I have encountered this myself.
I came across a variant of this in the census of a Carrier village carried out by Oblate priests, none of whom had much command of the language, in the 1870s. Several women are recorded as having been named tsandelh. What the priests didn't know is that tsandelh is not a name: it means "widow".