From Lognet, issue 92/3.

Sau La Sacdonsu

(From the Start-Giver = Founder)

The most important Institute development of the last three months, and one reported several places in this issue, is Dr. McIvor's success in finally recasting the machine grammar of Loglan—mainly the Preparser part of that system—so that our compound little words can now be correctly distinguished in speech as well as in writing from phrases composed of the same little words meant separately. Thus lepo vs. le, po, to take the classic instance of this problem, is one such case; but many more complex "compounding/decompounding" problems have been resolved as well. I'd put this achievement of our Cefli Talrultua (Chief Talk-Rule-Worker) on a par with Scott Layson's and my arriving, on a certain sunny afternoon in February 1982, on the top of the Corpus by finally writing a grammar that parsed it in one pass...a consummatory event which felt to us like reaching the top of a mountain. But in order to reach that summit, Scott and I had had to put certain "intractable" problems aside. Among them was the compounding problem, one that has remained pretty much as we left it for about ten years...until, in fact, Bob McIvor took it up again last Winter while I was still at sea. What he quietly set about doing for us was fixing a little-known defect of our parsing system: namely that it could handle only written input...that in fact it could not correctly distinguish compound little words from phrases composed of those same little words in the language spoken in the Loglandian streets.

Now it can. That is Bob McIvor's great achievement. And it will, I believe, prove to be an important one almost immediately. I expect Loglan to be asked to prove its mettle at that famous human-machine interface very interface, it now seems clear, where we are going to be mainly speaking and listening to our machines, not typing away at them and looking at their responses. What Bob McIvor has done for us—virtually singlehandedly, for the Keugru was called in merely to wonder at his work—is made us ready for that moment. And it doesn't matter in the least that his work has uncovered a few "stress-pause-punctuation" problems of a considerably higher order of difficulty than the "lepo/le, po" one; and that some of these remain unsolved.

One unsolved problem, as Bob will mention in his Sau La Takrultua in this issue, is that we're still not certain how best to speak strings of letter variables, or strings of variables preceded by connectives, so as to keep them distinct from keep u Nai (whether N) distinct from UNai ((the) UN), or Faize (Fe) distinct from the phrase Fai ze (F, ze), or the chemical symbol Cailei (Cl) distinct from the pair of letter variables Cai lei (C, l). We of the Keugru are confident that we can have all such problems solved by the time Lognet 93/1 appears next March. Well, most of them, soi spopa. And first-cut solutions.

* * *

Also in this issue is the longest Sau La Keugru we've ever published. Its length is partly due to the decisions we made in the Fall of '91 being held over until my return in May '92 "just to be sure". But part of its bulk was produced by a visit to The Institute by our cefli lodtua, Randolph Holmes, in August '92. We had a very fruitful evening, Randall and I, one in which we shaped up a number of proposals, some from him, some from me, which we then jointly presented to the Keugru. Most were adopted. These, too, are now part of our report. —JCB