This generally happens when the OneResponsibilityRule has been violated.
Example: An identifier for a class is given a very vague name because each meaningful name for the class is incomplete; it only describes part of the use of the class and not its whole responsibility.
Rather than fix the class with refactoring, the author takes the short cut and invents a name that doesn't clearly mean anything in order to avoid disinformation.
Can also happen in a well-factored class or method if the programmer is simply sloppy about naming things. I once came across a method called boolean NumVal?(String). Nuts. I had to search for the method's code before I found out that it returns true if every character in the string is a digit. I could think of a dozen better names for that method.
Why is NumVal? a bad name?
Does not describe what the method does. (It's not a verb, so how could it? -- See: NounsAndVerbs)
Does val stand for value or validate?
The method does not validate all numbers. In particular, it will reject numbers with decimal points or signs.
So it really checks for natural number (integer >= 0).
Except that there is no (practical) limit on the size of the number.
There may be valid reasons for the method to have these limitations, but the name of the method should reflect this.
If I saw a method called NumVal, I'd instantly expect it to return a numeric representation of the object whose method it is.