Institute of Formal and Applied Linguistics

Aravind K. Joshi (1929 - 2017)
Aravind K. Joshi (born August 5, 1929, in Pune, India, died December 31, 2017, in Philadelphia, USA), was the Henry Salvatori Professor Emeritus of Computer and Cognitive Science, a founding co-director of the former Institute for Research in Cognitive Science (IRCS) at the University of Pennsylvania and the recipient of numerous honors and awards (such as Honorary Doctorate of Charles University in Prague – 2013, Benjamin Franklin Medal in Computer and Cognitive Science of the Franklin Institute - 2005, Cognitive Science Society David Rummelhart Prize - 2003, ACL Lifetime Achievement Award - 2002, NAE Member - 1999, ACM Fellow - 1998, Founding Fellow - AAAI - 1990, IEEE Fellow – 1976), has been a member of the International Committee of Computational Linguistics since 1988.
The scope of his own research interests was very wide, covering the fields of Computational Linguistics, Cognitive Science and Artificial Intelligence, investigating also issues at the intersection of these fields and with extensions beyond them (cf. e.g. his long-time interest in macromolecular structures). He was a distinguished member of the international research community, appreciated for his intellectual curiosity and enthusiasm. He has been an inspiration for dozens of his PhD students and colleagues working all over the world.
It was a great honor and pleasure for all of us to have Aravind Joshi among us. He always acted as a thoughtful contributor to ICCL efforts to keep the COLING conferences alive. He always wanted to bring COLING to the United States again, and was very happy to see that finally, after many years, COLING 2018 is to be held there. Unfortunately, he will not see his efforts come to fruition in Santa Fe. We will miss him, not only at the conference, but we will miss, too, his mild, kind but always to-the point contributions to our discussions, his brightness and warm smile. We have lost a great intellectual. His departure leaves a huge empty place in our community.

Helmut Schnelle (1932 - 2015)
Helmut Schnelle received his first graduate degree in physics. During his work as a physicist at the Institute of Phonetics in Bonn he developed a deep interest in formal approaches to the study of language. His doctoral thesis in philosophy was on symbol systems, building on theories of Leibniz. The subsequent research for his habilitation (the continental European second doctorate) finally took him to linguistics: The thesis was entitled: "Prolegomena to the Formalization of Linguistics".
He was strongly influenced by but never fully satisfied with the logical foundation of linguistics. The search for alternatives to any purely symbolic formalization became the intellectual driving force behind his research during his entire career. Very early he was excited about dynamic networks as extensions to automata and then about neural nets as models of the language faculty. He also recognized the potential of computational modelling both for theoretical and applied language research.
1966 he was invited into the ICCL. 1986 he was appointed Full Professor of Linguistics at the Technical University of Berlin, where he soon assembled an entire community of linguists around him who were influenced by the "new" paradigms in the study of language such as categorial semantics, generative grammar, algebraic approaches, the New Prague School and computational language processing.
He was invited as guest professor to the Hebrew University of Jerusalem where he worked with Bar Hillel, and to MIT where he interacted with Chomsky and Halle. From 1976 till 1997 he was Professor for General Linguistics in Bochum. During this time, he concentrated increasingly on connectionist models of language learning and use. He tried to combine classical NLP algorithms with neural-net approaches and followed developments in psycholinguistics and the neurosciences closely. He became an influential and highly respected mentor and advisor for some of the most ambitious German and European research projects and centers. In Bochum he was member of the Research Center for Cognitive and Neural Networks (Kognet). 1989 he was elected into the Academia Europaea.
He became an emeritus professor in 1997 and moved to Berlin where he remained an active participant in conferences and other scientific events until his death in 2015.

Winfried Lenders (1943 - 2015)
Winfried Lenders studied philosophy, German, communication research and phonetics in Bonn, where he also received his PhD in 1970. In 1974 he successfully defended his habilitation thesis on communication research at the University of Bonn. Later in the same year, he was appointed Professor for Linguistic Data Processing at the same university. 1975 he founded the Association for Linguistic Data Processing in Germany, which after two transformations turned into today's Gesellschaft für Sprachtechnologie und Computerlinguistik (Association for Language Technology and Computational Linguistics). His central areas of research were corpus linguistics, machine translation and computational lexicography. In 1986, he organized COLING in Bonn and was invited to become a member of ICCL. 1993-97 Lenders served as President of the (German) Association of Linguistic Data Processing. One of Winfried Lender's last great achievements was the electronic edition of a large corpus of Immanuel Kant's writings. He was quite familiar with Kant's writings because Kant had played a central role in his PhD research. For several years, he worked very closely with philosophers on an electronic and printed Kant edition. He applied the methods of computational corpus linguistics and lexicography to improve the quality and readability of this edition considerably. His method also helped in the scholarly exploitation of the corpus for contemporary philosophical research.

Hozumi Tanaka (1941 - 2009)
Hozumi Tanaka graduated from the Tokyo Institute of Technology (TIT) in 1966 with a Master's degree in Control Engineering and immediately joined the Electro-Technical Laboratory (ETL) which had produced the first Japanese Machine Translation System (Yamato) and had been one of the centers of Computational Linguistics in Japan. He remained at ETL until 1983, working on topics including parsing, semantic analysis, and machine translation. He produced the LINGOL framework for NLP, based on LISP, that was in particular used by KDD (H. Sakaki) to build the KATE E-J MT system (1980-, used by Nippon Steel). He was a strong proponent of logic programming and was actively involved in the planning and execution of the Fifth Generation Computer Systems project (1982-1992), an ambitious attempt by the Japanese government to develop next-generation of "knowledge information processing systems". He received his PhD from TIT (1981) and accepted an Associate Professorship there (1983). He became a full professor in 1986 and remained in that position until retirement in 2005, when he took up a full professorship at a private university (Chukyo University). In 2009, he moved to the Japan Advanced Institute of Science and Technology as a Research professor.

During his time at ETL, TIT and Chukyo University, he supervised many young researchers and students in Computational Linguistics and became a father figure of the field in Japan. He was the technical lead on the Japanese government-funded CICC Machine Translation Project (1987-1995) between East and South-East Asian languages (Japanese, Chinese, Thai, Indonesian, and Malay). As a firm believer of the importance of knowledge and semantics in natural language processing, he designed a multilingual machine translation system for these languages based on a pivot language and on the previous ATLAS-II (by H. Uchida at Fujitsu, 1982-) and PIVOT (NEC, 1983-98) commercial MT systems. The project nurtured many computational linguists in East and South-East Asia. He was also one of technical leaders in another government project of Electronic Dictionary Project (EDR) from 1986 to 1995. The project was one of the first attempts of linking large vocabularies with concepts, based on corpus analysis. As of 2013, the EDR dictionary remains an important resource for Natural language Processing and is being kept up-to-date. Hozumi Tanaka joined ICCL at COLING 1994 in Kyoto, Japan.

Hiroshi Wada (1914 - 2007)
Hiroshi Wada joined Denki-Shikenjo (the Electro-Technical Laboratory) under the Ministry of Trade and Industry[ii] after graduating from the Department of Electrical Engineering, at the University of Tokyo. He was one of the first young researchers to be sent abroad by the Japanese government after WWII. He stayed at MIT for a year in 1951, where he realized the importance of the emerging technology of electronics, and especially that of the digital computer. Upon his return to Japan, he began a project to construct the first computer ever developed in Japan. The computer, ETL Mark III, was completed in 1956.

Hiroshi Wada understood that the essence of the digital computer was not merely in numerical calculation, but in a broad sense of computation. He led several ambitious projects at ETL, including Machine Translation, and the Optical Character Reader. In his conception, language could be the object of computation. The MT project introduced the first English-to-Japanese MT system (YAMATO) at the first Congress of International Federation of Information Processing (IFIP) in Paris, 1959.

Hiroshi Wada was a key figure in establishing the Information Processing Society of Japan (IPSJ) in 1960. In 1961, he started a special interest group for IPSJ in Machine Translation. The SIG evolved into the SIG on Natural Language Processing, and has always been the focal center of MT and NLP activities in Japan. The SIG organized a fact-finding tour of MT in the USA in 1965, and the members attended the first Coling held in New York. Hiroshi Wada joined ICCL as one of its founding members. Hiroshi Wada was a visionary in computer technology, and led the way for computational linguistics. He organized US-Japan seminars on MT, which attracted many young talents in Japan to the field. He and Makoto Nagao (another ICCL member from 1976) successfully hosted Coling 80 in Tokyo, which was an important milestone in the development of computational linguistics in Japan. After his tenure at ETL, he was invited to become a full-time professor by Seikei University (1964). He continued to be an active member of ICCL until the later years of his life. He always challenged young colleagues with his bold ideas.

Hiroshi Wada was forward thinking and visionary, and has been adored, revered and respected by many who have come to know him.

Olga Kulagina (1932 - 2005)
Olga Sergeevna Kulagina was a member of ICCL from 1969 and was one of the pioneers of machine translation and mathematical linguistics. She graduated from the mathematical faculty of the Moscow State University in 1954. Even before graduating, she was invited to a position at the Institute for Applied Mathematics of the USSR Academy of Sciences, where she worked her whole life. While a student, she became acquainted with Igor Melčuk, also a student at that time, who got her interested in natural language and its modeling, and formed her linguistic outlook. In March 1954, soon after the publication of the Georgetown experiment, Melčuk and Kulagina began working on a French-to-Russian machine translation system, the first in the USSR. They published their first results in 1956. Since then, machine translation became her lifetime project. Until the last day of her life she kept developing and enhancing her French-to-Russian MT system and working on general problems of machine translation. She summarized her vast experience in the book "Issledovanija po mashinnomu perevodu" ("Machine Translation Research") published in 1979.
Besides machine translation, she worked on mathematical modeling of natural language, especially in the early period of her career. Her paper "Ob odnom sposobe opredeleniya grammaticheskix ponyatij na baze teorii mnozhestv" ("On defining grammatical concepts on the set theory basis") published in 1958 has become a classic in the field. For political, and later for financial reasons, she could rarely travel outside of the Eastern bloc, but managed to attend most COLINGs until 1992. She was very active in French-Soviet cooperation, and, with Igor Melčuk, Alexander Zholkovskij, Victor Rosenzweig, Juri Apresyan, and Alexei Gladkij, contributed significantly to the theoretical foundations of the first Russian-French system built in Grenoble by Bernard Vauquois and his CETA team.

Antonio Zampolli (1937 - 2003)
Computational Linguistics started in Italy with Antonio Zampolli. He held the first chair of Computational Linguistics and founded the Istituto di Linguistica Computazionale of the CNR in Pisa that now carries to his name.

In the early '70 he organized the famous Pisa Summer Schools, at a time when summer schools were not trendy, bringing to Pisa the greatest NLP/CL names of the time. He always remembered that Joan Bresnan and Ron Kaplan gave birth to lexical functional grammar on a tower in San Gimignano. These schools have formed a whole generation of computational linguists all over Europe.

He joined ICCL in 1973, organised the 5th COLING in 1973 in Pisa and was chair of two other COLINGs. He founded many associations, boards, committees, networks (EURALEX, ELSNET and ELRA to mention just a few) and was president almost of everything.

He was a man of great vision, capable of anticipating and creating the future, always pushing towards new directions and initiatives, often struggling to make his intuitions a reality. He also had the capability to mix people from different communities, thus creating perfect mixtures to develop new ideas. He was able to communicate to everyone his enthusiasm and his passion.

He 'invented' the field of Language Resources, at a time when it was almost a shame to speak about language data. But he believed in them and he pushed for their development. The beginning of this data era was the Grosseto Workshop in 1985. Now everyone speaks about language resources, as if they are normal tools, but he had to fight to open this direction of work. He created a long series of standardisation initiatives, from TEI to EAGLES, and innumerable projects to set up a European infrastructure for language resources, culminating with the setting up of a Network of National projects (ENABLER). He also set up LREC, a highly successful conference, where again there is a biennial gathering of many different communities.

He was a 'maestro' for so many of us, a 'living legend' as has been said. But also a man who was loved, for his humour, his sympathy, his enthusiasm, his vitality, his intuitions, his friendliness to all, and his wonderful stories. Antonio Zampolli continues to live with all who continue to work in one of the many directions that he opened.

Hans Karlgren (1936 - 1996)
Hans Karlgren was an unusual personality. He combined a noble character with creativity and organizational talent, and he was one of the pioneers in the early days of our field. Born in Stockholm in 1936, Hans worked as shorthand reporter for the Swedish Riksdag from 1951 to 1964. He was the editor of SMIL (Statistical Methods in Linguistics) from 1960 to 1972 and President of KVAL Institute for Information Science from 1960 until his death. From 1965 he worked as authorized translator from Finnish to Swedish, and from 1969-1996 he was a member of the committee on Linguistics in Documentation of the International Federation for Information and Documentation (FID/LD). After his PhD thesis on Categorial Grammatical Calculus, he became in 1974 a lecturer and associate professor in General Linguistics. From 1983 he was a member of the Expert Committee on translation of Finnish literature into Swedish. He died suddenly and unexpectedly on December 31st, 1996.
The writings of Hans Karlgren include articles on General and Computational Linguistics, with a particular emphasis on quantitative problems and information science. He also wrote on stenography as well as on political topics and on issues of current language use in Sweden, most of them appearing in the daily newspaper Svenska Dagbladet.

As a founding member of the ICCL, Hans Karlgren played a formative role in its successful development and of Computational Linguistics in general. In 1969 he organized the third COLING at Sanga Säby near Stockholm. In 1990 he served as program chairman at the 13th COLING in Helsinki, where he is said to have read every submitted paper himself. Hans organized other conferences and workshops in Computational Linguistics such as the FID/LD Workshop on Linguistics and Information Science in 1976.

David Hays (1928 - 1995)
David Glenn Hays may be considered the father of the field of computational linguistics, not only by inventing its name, for he was also the force behind the foundation of the Association of Computational Linguistics and the International Committee on Computational Linguistics. He contributed substantially to a range of the disciplines: machine translation (being involved in the development of some of the first machine translation systems), formal linguistic theory (with his seminal paper on dependency grammar), corpus linguistics (he participated in the earliest efforts in that field), in addition to his deep interest in social and evolutionary systems. David Hays received his education and PhD (1956) in sociology at Harvard University, and worked as a social scientist at the RAND Corporation (1955-1968). He then became a Professor of Linguistics, of Computer Science, and of Information and Library Studies at the State University of New York at Buffalo (1968-1980), where he chaired the Department of Linguistics from 1968 to1970 and served as the Director of the Linguistic Institute in 1971. He received many honors and awards for his scientific contributions, he was the Founding Chairman of the International Committee on Computational Linguistics (1965-1969) and served as the President of the Association for Computational Linguistics in 1964.

Don Walker (1928 - 1993)
Don Walker over three decades decisively influenced the development of Computational Linguistics, both by his scientific contributions, as well as by his commitment to the implementation and support of scientific meetings, publications, acting as secretary of different societies, and as a member of the ICCL. He was a talented visionary and mediator, and a promoter of numerous young scientists (see Donald E. Walker: A Remembrance by Barbara Grosz and Jerry R. Hobbs, in Computational Linguistics March 1994).

After his study of linguistics and psychology in the early 1960s Don became the head of a research group on computational grammar at The Mitre Corporation, Bedford, MA, with the particular concern of building a transformational parser for computational applications. In 1973 he started the Natural Language Processing group at SRI and developed it into one of the leading research groups in this field in the world. Under his direction early ARPA/DARPA-funded speech and language projects were conducted. He created the concept 'lexical knowledge base' and was one of the first to recognize the importance of natural language corpora as resources for future computational linguistics research, a development which was at the centre of his activities as Director of the Artificial Intelligence and Information Science Research Group at Bellcore (Bell Communications Research), Morristown, NJ. He inspired the worldwide collection of language data, which resulted in the ACL Data Collection Initiative and the Text Encoding Initiative. The legendary Grosseto workshop On Automating the Lexicon, 1986, was mainly initiated by him and can be seen as what Calzolari and Zampolli "marking the starting-point of a new phase in the field of computational linguistics. Don was heavily involved in the creation and management of several important organizations in Artificial Intelligence and Natural Language Research.

He was the driving force behind the 1962 foundation of the Association for Computational Linguistics, whose Secretary-Treasurer he was from 1976-1993. With his help the European Chapter of the ACL (EACL) was formed 1982. From 1980 through 1993 Don Walker served as Managing Editor of the journal AJCL/CL, and he was also for a long time (1969-1993) Secretary-Treasurer of IJCAII. He became a member of the ICCL in 1984 at COLING 84 in Stanford where he was committed himself to a fruitful relationship between ICCL and the ACL. Don died in 1993 after a long battle with cancer at the peak of his activity. In his memory, the ACL Don and Betty Walker International Student Fund and the IJCAI Donald E. Walker Distinguished Service Award were donated.

Pieter Verburg (1905 - 1989)
Pieter Adrianus Verburg studied classical languages at the Vrije Universiteit, Amsterdam. During this time he also spent a semester in Freiburg and another in Berlin. After completing his coursework, he worked in London as a private tutor while preparing a dissertation on metaphor, which remained unfinished. He then returned to the Netherlands to become a classics teacher in Wageningen. In World War II he was active in the Dutch resistance movement. After the war, Verburg became Director of Het Nationaal Instituut. On 30 November 1951 he received his doctorate at the Vrije Universiteit for a dissertation on Language and Functionality. In 1957 Verburg became Professor of General Linguistics at the University of Groningen, where he taught linguistics and later also philosophy of language until his retirement in 1975. During his scientific career he mainly worked on the origins and evolution of central notions in linguistic theory, combining reviews of selected chapters in the history of European linguistics with critical reflection. In 1969 at the third COLING in Sanga Säby, Verburg reported on an early forerunner of a statistical view of human language processing in Hobbes' philosophy of language. He turned this COLING talk into a journal article titled "Hobbes' Calculus of Words" published 1970 in Statistical Methods in Linguistics. In later work, he touched on fundamental issues of modern linguistics, for instance when critically discussing Leibniz' linguistic system and reflecting on conflicting constructions of the study of language as ars or scientia.

Guy Rondeau (1930 - 1987)
Guy Rondeau was an ICCL member from 1976 until his untimely death at the age of 57 in 1987. He was a pioneer of computational linguistics (CL) in Canada, and more especially MT. He was from Quebec and his first language was French. He began his university career at the University of Montreal (UM) as a specialist in second-language teaching. His mentor was a well-known French teacher of linguistics and translation, Jean-Paul Vinay, and he did his doctorate in France, where he retained strong connections. In the 1960s he lived through the period of intellectual renewal in Quebec society known as the Quiet Revolution. It opened up new possibilities for Quebec's universities and their academics. Thus Rondeau became interested in CL and went to see what was going on at MIT. He made a French translation of the manual for Victor Yngve's language COMIT and organised a CL section at UM called CETADOL.

In 1965, the National Research Council of Canada launched an MT R&D project for translation from English to French. They gave contracts to several teams, and one of them was CETADOL. Rondeau forged a strong link between CETADOL and Bernard Vauquois' team at the University of Grenoble and he recruited a brilliant French computer scientist, Alain Colmerauer, who provided CETADOL (renamed TAUM) with a powerful programming language for MT called Q-systems (Systèmes-Q). After that, Rondeau quit UM and MT and finished his career at Laval University in Quebec City as an influential professor of terminology theory and practice.

Rondeau's strengths lay in his abilities as a far-seeing organiser, talent spotter, and fund raiser. He chaired very effectively the organising committee for COLING 1976, which was held in Canada.

Bernard Vauquois (1929 - 1985)
Bernard Vauquois' initial training was in mathematics and astronomy. He became interested in computer science while working on his main thesis (on astronomy), around 1955, when the field came into being in France. For his second thesis, he was asked to explore (by then quite new) questions of decidability and undecidability. This is how he was introduced to the fundamental domains of logic and formal languages. Enthusiastic about these new horizons, he accepted Professor J. Kuntzman's proposal to come to Grenoble and launch a curriculum in computer science and formal languages, as well as to start a research laboratory on MT (CETA-G, branch of CETA-P, that stopped in 1962). This was in 1961. He began by touring the world, gathering ideas from and starting cooperation with the best researchers of the time (in Europe, in the US, in the USSR and Eastern countries, and in Japan).

He was one of the founders of the ICCL in 1964-65, and its President from 1967 (when he organized the second COLING in Grenoble) to 1984, passing on the role to Martin Kay (who organized then the first COLING-ACL in Stanford). He was a charming man, a visionary scientist, and an excellent organizer with whom everything seemed simple. His death was deeply regretted all over the world, and particularly by his ICCL colleagues. For more than 25 years, until his sudden death in October, 1985, Bernard Vauquois was very active as researcher in MT and professor in statistics, and then logic, formal languages and NLP. In MT, he was a real pioneer, and continuously contributed decisive innovations.

Let us summarize them here.

1960-65: enlightened application of the theory of formal languages (analysis by attributed CNF grammars and optimized CKY, production of a best dependency tree, abstraction to a hybrid semantic).
1965-70: integration of modern linguistic theories in the first ever full-2G MT system, and of a metalanguage for transformational grammars (1967).
1970-75: use of modular procedural grammars and heuristic methods to raise quality on sublanguages like those of technical manuals or standardized abstracts. Ariane-78 system, with an in-built revision facility (MT for revisors). Shift from the analysis to the transduction paradigm and enlargement of translation units to several paragraphs or even pages.
1975-80: introduction of "multilevel" structural descriptors, encoding several levels of linguistic interpretation on a semi-concrete lexicalized and decorated syntagmatic tree, as well as some ambiguities and advices or orders to subsequent processing phases.
1980-85: introduction of the "static grammars", a semi-formal declarative specification language for the correspondences between texts and structures (Chappuy & Vauquois 1983). That led to an efficient technology for building rule-based lingware components.
1983-85: towards a new treatment of ambiguity. Vauquois worked on adding to the "construct boards" of the "static grammars" "ambiguity boards" that are preference rules between two structure patterns, for the same string pattern.

Alexander Ludskanov (1926 - 1976)
Alexander Ludskanov was a Bulgarian semiotician. He was a member of ICCL from about 1968 until his untimely death at the age of 50 in 1976. He came from a family that was prominent in Bulgarian politics, and he learnt French and Russian as a child. But his life was upturned by the Communists' rise to power after the Second World War. He had to do a period of hard labour, and this seriously undermined his health. It was his knowledge of Russian that subsequently enabled him to find work as a translator for a Bulgarian-Soviet friendship magazine and to return to the University of Sofia. It was then that he became interested in the general theory of translation and did his thesis on it. Again his knowledge of Russian helped him, through his contact with Soviet linguists. In 1964, in imitation of the Russians, the Bulgarian Academy of Sciences set up a Bulgarian-Russian MT project at the Institute of Mathematics with Ludskanov as director. He was happy to remain there for the rest of his life because, as he said, "Mathematics is the only department of academia that they can't politicise."
Everybody who heard him lecture was impressed by the logic of his expositions. His magnum opus, published in Bulgarian in 1967, was his book "Human and Machine Translation." It was translated into French (by himself), German and Polish during his lifetime, but unfortunately not into English. He sought to formulate a general theory that would unify all the genres of human translation as well as machine translation. For this, he regarded translation as a transfer of signs through a mental form of them. These signs represented not only linguistic input but also all the "information necessary for translation." He proposed as well a model of the history of translation based on text types. The computational component of the book is now out of date, but it is still of value for its translation theory, and this latter part has been published quite recently (2008) in Italian.