True, but you still have to build up a forest from which every parse can be extracted (as Alpino does). Of course, it does reduce the cost of ambiguity. E.g., I

True, but you still have to build up a forest from which every parse can be extracted (as Alpino does). Of course, it does reduce the cost of ambiguity. E.g., I implemented packing in the Alpino chart generator, and there are very many edges with the same 'semantics' that can be packed, especially since Alpino allows for a lot of optional punctuation, and there are often roots plus subcategorization frames that give multiple inflections. With a beam search, unpacking is a lot more efficient. There are also many other possible optimizations. E.g. with a left-corner parsers you can exclude left-corner spines that will probably not lead to a probable parse (Van Noord, Learning Efficient Parsing, 2009, and other work). And, of course, reduction of lexical ambiguity by restricting frames using a part of speech tagger. Still, even in this best-1, optimized scenario, real-time parsing of long sentences is still hard. So, when parsing large corpora we usually apply time and space limits (which is easy to do in Sicstus Prolog, with good recovery). Thanks for the link to the paper!