The article explains it in terms of teaching methods: it sounds like the method they describe involves a consistent relation of the order of strokes to the parts of the word being recited with it; that would make the phonetic associations (even though the symbology is not phonetic) consistent. A consistent phonetic writing system (which is likewise less likely to reveal dyslexia) has a similar feature. English is taught phonetically, but has intense phonetic irregularities, so the phonetic connections at present but ambiguous. If phonetic associations with the symbol through learning have an anchoring effect, that would seem to explain the identified effect.