Identifying function words

[Stephen.Bax]

Grammatical structures can be quite varied and require us to analyze longer chunks of text. It's also harder to know what success would look like. Even if we managed to model the grammar of the text how would we prove that we had the right solution? Knowing the grammar alone would still need a further step.

But your way of thinking is quite right. We can map language universals onto the text and gain some kind of insight. This is what I've been doing looking at the structure of words trying to find out how the apparent structure reflects sounds. The great thing is that we can analyze words alone, have a much tighter understanding of the whole sound system, and the outcome is, potentially, a readable text. If we learnt enough about the different characters, how they're structured, how they interact, and what sounds they can and cannot be, the whole of the text will be open to us and we can simply say 'yes, this is a known language', or 'no, this is gibberish'.

The really, really, great thing is that each step can be built without reference to any particular language. We supply a sound linguistic argument for each conclusion and at no point waste time trying to fit the text to our preferred solution. (Well, excepting of course the assumption of a linguistic solution!)

Has this approach ever been successful in decoding any previously unknown language or script?