That problem is known. The problem is that XLE does, by default, not use UTF-8 (neither when reading the LFG file nor when reading the sentence). This means it interprets the "ü" as something different both in the grammar and in the sentence (namely the "ü"), but at least it parses.
You can add a line to the grammar to tell XLE it should interpret it as UTF-8, but then you'd also have to tell XLE to interpret the input as UTF-8. That can probably be done, I'm working on it in xle-docker, it just isn't as easy as I thought, because the encoding does not apply to the xlerc file in which I save the sentence to be parsed.
Jakob Moser authored