From cbmvax!uunet!PICA.ARMY.MIL!protin Fri Mar 29 14:06:26 1991 Return-Path: Date: Fri Mar 29 14:06:26 1991 From: "Arthur W. Protin Jr." (GC-ACCURATE) To: lojban-list@snark.thyrsus.com Subject: Re: Response to jack Waugh on 'hard'cmavo (old posting) Message-Id: <9103291213.aa12257@COR4.PICA.ARMY.MIL> Status: RO Folk, I feel compelled to comment on something, so here it is: On Tue, 5 Mar 91 10:47:08, uunet!mcnc.org!aurs01!aurw31!waugh (Jack Waugh) wrote: >An adequate model to use for the first stages of listening would be to >say that the listner parses first and then starts semantic processing. >Parsing is casting the utterance into a tree, so as to know the grouping >relationships. Once this is done, the listener turns to the semantics >of the words. ... I disagree strongly! While that model may be fine for detailed analysis, the model for realtime comprehension is more like a dictionary lookup for each word (the dictionary may change as a result of the previous word) and the little "n", "vt", "vi", "adj", etc. being used to perform an increment parse interleaved with the construction of a complex semantic evaluation. My studies have convienced me that children privately and personally reinvent syntax to permit them to deal with utterances that have meanings that differ from the mere juxtapositoning of simple meanings of the components. It seems that this development is the alteration of the meaning associated with the token (word) to include some form of restriction about how this meaning may combine with other meanings (more than what the idea alone would require). For example the idea for the word "carry" gets modified such that it is more important to the understanding which agent preceeds the word than is the relative strengths and weights of the agents. Thus initially the child will combine a reference to a child, a parent, and "carry" such that the parent will always carry the child and only after the modification will find humor in the suggestion that the child should carry the parent. It is only after modifying a significant portion of a child's vocabulary that the child will both transfer this new property to other tokens (in complementary form as in "carry" implies an agent), and generalize tokens as having some structure requirements (and expect them in new vocabulary and incorporate them at introduction). What we call syntax is simply the abstraction of the effects of those binding and combining requirements attached to each word of our collective vocabularies. And because of our limits of representation and manipulation of ideas, we further modify those restrictions so as to permit an infinite (or nearly so) collection of instances to be represented by a (relatively) small set of rules. I believe the loglan project receives some additional impetus from our need (strong desire?) to simplify and reduce those rules even further. thank you all for this opportunity to speak, Arthur Protin Arthur Protin These are my personal views and do not reflect those of my boss or this installation.