I'd like to know why Spanish was only 85% for reading in this study. Did they use Spanish texts that were full of English words or foreign names or something li

I'd like to know why Spanish was only 85% for reading in this study. Did they use Spanish texts that were full of English words or foreign names or something like that? Or did they consider some informal colloquial pronunciation to be the "correct" pronunciation? Or did they have a confusing mixture of different varieties of Spanish in their dataset? I'm finding it hard to think of frequent cases in which it is incorrect to pronounce "pure" European Spanish exactly as it's written. Another possible explanation: perhaps their ANN model was incapable of working across word boundaries? (The "n" in "en Bolivia" is pronounced like "m": it's perfectly regular and a simple rule but you do have to look beyond the current word.) EDIT: When I write "exactly as it's written" I mean, of course, that you compute the pronunciation from the spelling by applying some simple rules. For example, B and V are pronounced the same way, and that way depends on the context, but the rules are very simple: they can be expressed in a few hundred words. Also, I don't consider an overly formal pronunciation to be "incorrect" for the purposes of this discussion!