William Gale

From Curriepedia
Revision as of 16:12, 5 August 2018 by Admin (talk | contribs)
Jump to: navigation, search
William Gale, circa 1997

William A. Gale (1939 or 1940 - 6 August 2002) was a scientist and core member of the First Millennial Foundation in the 1990s.

Early life

Gale grew up in Houston, Corpus Christi, and Fort Worth, Texas. His B.A. in math is from Rice University, as is the Ph.D. in physics.

Gale worked for several years for a company called Bellcomm which was a Bell System company set up to consult for NASA manned space flight during the Apollo years. With the decline of the manned space flight program, Bellcomm was absorbed back into Bell Labs and Gale followed it. Over the years, Gale published in physics, radio astronomy, bioastronomy, economics, computer science, applications of artificial intelligence in statistics, and computational linguistics.

Gale founded the Society for Artificial Intelligence and Statistics, published some five dozen papers, and a half-dozen books.

In the mid-1990s, Gale retired due to disability. Gale could not leave his house without help but he was "delighted to see people. Please drop by if you are in the New York area. My house is about 40 minutes from Manhattan. I've read a lot of science fiction, played board war games since Tactics II, and shifted to computer strategy games (but especially Go). I've written some poetry. I enjoy classical music--especially Beethoven chamber music."

Gale "wrote a complete operating system for the Apple II starting from a very simple bootstrap self-compiling compiler, moving to an assembler and a self-compiling compiler-compiler written in the bootstrap language, a C-level language, a file system, shell, editor, etc. What a trip that was! I had so much fun that I decided I should find a way to be paid for it, and that's when I picked up the AI applications in statistics at work."

Gale "grew up without TV and rarely watch it, so I'm kind of not in the current culture. However, I watch a few videos now and then. I collect videos of opera, modern dance, musicals, and outstanding movies--things worth seeing again and again. My favorite movies: Seven Samurai, Fantasia, Ordet."

In 1997, Gale's "ex-wife, Lynn, lives across the street with our 14-year-old daughter, Rebeckah; our 27-year-old daughter, Marion, lives with me. They all are helping me with things I can't do any more. Lynn practices law. Marion is deeply into the Society for Creative Anachronism. Rebeckah hangs out on the phone."

As a linguist

Gale, as he liked to be called by his friends and family, had extremely broad interests, both professionally and otherwise. His professional career at Bell Labs included radio astronomy, economics, statistics and computational linguistics. He always had lots of collaborators because he was such a joy to work with:

I only ever really knew Bill via e-mail, yet he represented one of the most significant academic contacts I’ve ever had. The Good-Turing paper that we collaborated on was definitely one of the most fulfilling high points of my publishing career – apart from any intellectual value it may have, it was more fun than almost anything else I’ve worked on in my professional role. It was a privilege for me that Bill accepted me as a partner on that task.
— Geoffrey Sampson

Gale had a remarkable tendency to jump start research areas in computational lin-guistics (and elsewhere) that would later become extremely popular. Some examples include:

  • Parameter estimation
  • Word sense disambiguation
  • Parallel corpus alignment and text classification
  • Lexical statistics

Always curious, and always a spectacular teacher, he would start his collaborator going on some new topic that would keep them busy for years to come, while he wandered off to create yet another new research area. This special issue (and the special interest group SIGLEX) is largely an outgrowth of the year or two that he spent having a good time playing with word senses in the early 1990s. There was alot to learn from him, both as a scientist and as a person.


He died 6 August 2002.

Following his death, a special issue of Natural Language Engineering was dedicated to him. [1]

He left a bequest to the FMF to fund Richard Crews' Space Environments Ecovillage (SEE) project. It was one of the only significant donations FMF ever received.


Parameter estimation

Church, Kenneth W. and Gale, William A. (1989) Enhanced Good-Turing and Cat-Cal: Twonew methods for estimating probabilities of English bigrams.Second DARPA Workshop onSpeech and Natural Language,CapeCod.

Gale, William A. and Church, Kenneth W. (1990) Poor estimates of context are worsethan none.Proceedings of DARPA Speech and Natural Language Workshop,Hidden Valley,Pennsylvania.

Church, Kenneth W. and Gale, William A. (1991) A comparison of the enhanced Good-Turingand deleted estimation methods for estimating probabilities of English bigrams.ComputerSpeech&Language5(1): 19–54.

Gale, William A. and Church, Kenneth W. (1994) What’s wrong with adding one? In:Oostdijk N. and de Haan P., editors,Corpus-Based Research into Language: In honour ofJan Aarts,pp. 189–200. Amsterdam: Rodolpi.

Gale, William A. and Sampson, Geoffrey (1995) Good-Turing frequency estimation withouttears.J. Quantitative Linguistics2(3): 217–237.

Church, Kenneth W. and Gale, William A. (1995) Poisson mixtures.Natural LanguageEngineering1(2): 163–190.Church, Kenneth W. and Gale, William A. (1999) Inverse document frequency (IDF): Ameasure of deviations from Poisson.Proceedings Third Workshop on Very Large Corpora,pp. 121–130.Wordsense disambiguationGale, William A., Church, Kenneth W. and Yarowsky, David (1992) A method for disam-biguating word senses in a large corpus.Computers & Humanities26:415–439.Gale, William A., Church, Kenneth W. and Yarowsky, David (1992) Estimating upper andlower bounds on the performance of word-sense disambiguation programs.Proceedings30th Annual Meeting of the Association for Computational Linguistics,pp. 249–256. Newark,Delaware.Gale, William A., Church, Kenneth W. and Yarowsky, David (1992) One sense perdiscourse.Proceedings of the DARPA Speech and Natural Language Workshop,pp. 233–237.New York.Gale, William A., Church, Kenneth W. and Yarowsky, David (1992) Using bilingualmaterials to develop word sense disambiguation methods.Fourth International Conferenceon Theoretical and Methodological Issues in Machine Translation,pp. 101–112. Montreal.Gale, William A., Church, Kenneth W. and Yarowsky, David (1995) Discrimination decisionsfor100,000 dimensional spaces.Ann. Operations Res.55:323–344.Parallel corpus alignment and text classificationGale, William A. and Church, Kenneth W. (1991) A program for aligning sentences in bilingualcorpora.Proceedings of the 29th Annual Meeting of the Association for ComputationalLinguistics.Berkeley, CA.Gale, William A. and Church, Kenneth W. (1991) Identifying word correspondences in paralleltexts.Proceedings of the DARPA Speech an Natural Language Workshop.Church, Kenneth W. and Gale, William A. (1991) Concordances for parallel text.Proceedingsof the Seventh Annual Conference of the UW Centre for the New OED and Text Research,pp. 40–62. Oxford.Dagan, Ido, Church, Kenneth W. and Gale, William A. (1993) Robust bilingual wordalignment for machine aided translation.Proceedings of the Workshop on Very LargeCorpora,pp. 1–8. Columbus, OH.Lewis, David D. and Gale, William A. (1994) A sequential algorithm for training textclassifiers.Proceedings 17th Annual International ACM-SIGIR Conference on Research andDevelopment in Information Retrieval,pp. 3–12. Dublin, Ireland.Lexical statisticsChurch, Kenneth W., Gale, William A., Hanks, Patrick and Hindle, Donald (1989) Parsing,word associations and typical predicate-argument relations.International Workshop onParsing Technologies,CMU. Dedication to William A. Gale277Church, Kenneth W., Gale, William A., Hanks, Patrick and Hindle, Donald (1991) Usingstatistics in lexical analysis.Lexical Acquisition: Exploiting Online Resources to Build aLexicon,pp. 115–164. Hillsdale, NJ: Lawrence Erlbaum.Church, Kenneth W., Gale, William A., Hanks, Patrick, Hindle, Donald and Moon, Rosamund(1994) Lexical substitutability. In: Atkins B. T. S. and Zampolli A., editors,ComputationalApproaches to the Lexicon,pp. 153–177. Oxford University Press.Tzoukermann, Evelyne, Radev, Dragomir R. and Gale, William A. (1995) Combining linguisticknowledge and statistical learning in French part-of-speech tagging.EACL Workshop onVery LargeCorpora.Dublin, Ireland.Sproat, Richard, Shih, Chilin, Gale, William A. and Chang, N. (1996) A stochastic finite-stateword-segmentation algorithm for Chinese.Computational Linguistics22(3): 377–404.Kernighan, M. D., Church, Kenneth W. and Gale, William A. (1990) A spelling correctionprogram based on a noisy channel model.Proceedings Thirteenth International Conferenceon Computational Linguistics,pp. 205–210.


Natural Language Engineering 8(4): 275–277.c ©2002 Cambridge University PressDOI: 10.1017/S1351324902003066 Printed in the United Kingdom [2]

Autobiography from FMF website, 28 February 1997 [3]

New York Times, 16 February 1978 [4]

Dr. William A. Gale of the Bell Telephone Laboratories carried the colonization concept to its ultimate extreme. He envisioned its extension throughout the Milky Way galaxy, formed of more than a hundred billion stars, and then possibly to other galaxies.

Such dynamic expansion must be expected of any intelligent life form, he said, transforming a galaxy and perhaps even a cluster of galaxies. Since there is no evidence of any such transformation nearby, he suggested that the nearest intelligent life form must be very distant.

Dr. Gale, a physicist, served with Bellcomm, the Bell system's contribution to the Apollo moon‐landing project, before joining Bell. Telephone Laboratories. He envisioned an average population growth of half a percent annually, leading to a billionfold increase in a few thousand years.

This would saturate the solar system and migrations to other stars would become necessary. By then, he suggested at a press briefing before the meeting, technology would be able to develop habitats to orbit a star regardless of whether Earth‐like planets exist there. For such a construction, he added, “we will take whatever is available.”

In this way, the galaxy could be fully colonized in one to 10 million years—a short time, geologically speaking, ‘ and comparable to that in which the human race has inhabited the earth. If travel speeds in excess of one‐tenth the speed of light can be achieved, he said, colonization of other galaxies will become .

possible. He suggested that a lookout be kept for galaxies that seem to be vanishing as their stars are exploited. Skeptics in the audience commented, however, that his proposals ignored the difficulty, if not irrationality, of prolonged space travel.
— New York Times, 16 February 1978
We are sad to report that William A. Gale, statistician and computational linguist, passed away at his home in Maplewood, New Jersey on August 6, 2002.

He was 63. Bill grew up in Houston, Corpus Christi and Fort Worth, Texas, and earned his Bachelor's and Ph.D. degrees in physics from Rice University. He started work at AT&T Bell Labs doing radio astronomy for the Venus probes.

In 1980 Bill moved to the statistics department at Bell Labs, where he collaborated with economists, and started to work more and more on issues related to computer science. With Daryl Pregibon he built the first demonstration of the feasibility of applying ideas from expert systems in statistics. In the late 1980's Bill's interests turned to computational linguistics, and the statistics of language. He worked initially with Ken Church on problems of term-frequency distributions, but was soon collaborating with many natural language researchers at Bell Labs and beyond.

Bill was enthusiastic about every problem he tackled, and his enthusiasm and his skill in applying mathematical models to novel domains will be sorely missed. We will also miss his patience in helping non-mathematicians understand how to apply statistical models to linguistic problems.

Finally, and perhaps more than anything, we will miss his gentility. As anyone who has worked with Bill will know, it was always a pleasure to talk to him, whether about technical matters, or about the wider realms of life.

- Ken Church, Chilin Shih, Richard Sproat
— Richard Sproat <rwsresearch.att.com>, 8 August 2002 [5]