Hi, Brian Provinciano recently pointed out to me that word classes, if they are really described by a sequence of 12 bits in vocab.000, don't really limit themselves to using the lower 8 bits. While examining this, I found out that FreeSCI uses an incorrect internal representation for those classes (the four highest bits are in the least significant nibble, with the 8 lower bits are shifted 4 bits to the left; still, the parser appears to work for some reason). I examined the words used in SQ3, and found a few that had some of the higher bits set: * 4ff about 800 and 400 any 4ff anyword 4ff but 400 for 800 if 400 nor 400 for 400 until 800 while 800 with 800 without 800 The standard parser grammar does not come with any rules to handle words of classes 0x800 or 0x400, so I'm wondering whether those got any special treatment.... any ideas? Also: Are there any volunteers for fixing the vocab bug(s)? llap, Christoph
