I am deeply honored to receive the ACL's Lifetime Achievement Award. I'm especially honored when I look back at the list of previous winners—Chuck Fillmore, Eugene Charniak, Eva Hajičová, Fred Jelinek, Martin Kay, Aravind Joshi, and the others—they're all my heroes.I was of course delighted to learn of this award. The most we can hope for in life is to take part in the conversation, and an award like this means that you've taken part in the conversation.It seems to be a tradition to begin with a few formative anecdotes from childhood. For me, it all begins before I was born. My grandfather, a crusty old country lawyer in southern Indiana, told my father not to bother trying to go to law school. “You don't know English grammar,” he said. “You'll flunk out.” My dad accepted the challenge, bought a book entitled English Grammar, by Smith, Magee, and Seward (1928), and mastered it. He went on to become a very successful lawyer.Fast-forward to when I was in junior high school. My dad was distressed that my English classes looked to him more like social studies, and barely touched on grammar. So he persuaded me—actually, he probably bribed me, but I can't remember what with—to master that same book, English Grammar by Smith, Magee, and Seward. This was a concession, because I was a math nerd, reading only textbooks on trigonometry and calculus, as my way of avoiding the humiliation of playing baseball. But I read the book, and I was amazed. English grammar was just like math! It had the same sorts of rules, the same kinds of abstractions, the same types of puzzles. It was actually fun!In my junior or senior year of high school we had to take something called the Kuder Preference Test, which would help us decide what career to choose. I scored high in math and in language. So my high school counselor told me I should write math books. In fact, she got it exactly backwards. It wasn't that I should do language about math. It was that I should do math about language.I've met any number of computational linguists with a similar story. They grew up not knowing whether they wanted to be a physicist or a poet. They just knew both sounded fascinating. Then they discovered our field.My last near miss happened the week I was drafted into the Army. They gave us a battery of aptitude tests to see what specialties we'd be best for. One of the tests was to see if we should be sent to the Monterey Language School. Looking back on it, I realize now it was testing how well you could understand formal language theory. They'd give you a bunch of rules for an artificial language, and you'd have to say whether different strings were in or not in the language. I'd never seen anything like it, but it was really fun to do. Later I met with a personnel specialist who went over my test scores. I got a 46 out of 50. He ignored that until I pointed it out to him. Then he said, “That's a mistake. Nobody ever gets more than 6 or 7 points on that test.”I said, “No, I think it might be correct.”He said, “It doesn't matter. You're not going to the Monterey Language School. You're going to South Vietnam.”Actually, I didn't go to the Monterey Language School or to South Vietnam. I spent two years in South Carolina, and was glad to be there. How I managed that is a story for another occasion.So I didn't really discover computational linguistics until my third year in graduate school at New York University. In October I passed my oral exam in topics like algebraic topology and complex analysis, by one generous yes and two abstensions. In the subsequent months I discovered more and more facts about myself—for example, that I was never going to figure out a faster way of multiplying matrices, and that fascinating though recursion theory might be, I was never going to prove a theorem that Hartley Rogers would be compelled to include in his next edition. As I surveyed vaguely plausible fields, I realized I had no idea what the next problem to solve would be or even what makes a problem interesting.Then in April, when I had nearly resigned myself to becoming a taxi driver, I discovered New York University's best-kept secret: Naomi Sager's Linguistic String Project. I think it is also computational linguistics' best kept secret as well. She was motivated by the science, not by the performance, and her very impressive work is nowhere near as well-known as it should be. I think her Linguistic String Grammar (Sager 1981) ranks, as a computational specification of English syntax, with Pollard and Sag's Head-driven Phrase Structure Grammar (1994), for thoroughness, insight, and elegance. So, for example, in 1992 when we developed the FASTUS system for information extraction using cascaded finite-state transducers (Hobbs et al. 1997), it was straightforward to copy the rules for Noun Groups straight from her grammar. It's no accident that in the late 1980s during the Strategic Computing Initiative and in the early 1990s in the Message Understanding Conferences, three of the most important efforts were led by Linguistic String Project alumni—Ralph Grishman's group at New York University, Lynette Hirschman's at Unisys, and my group at SRI International. I think the most important lesson I learned from Naomi Sager was to look closely at the data and to take it seriously.My other thesis advisor was Jack Schwartz. He was a polymath, so to speak. I took a course in logic from him. I knew about his book on compilers and the classic Dunford and Schwartz on functional analysis. But when I saw his book on mathematical economics and his book on the theory of relativity, I did some research to see if there was more than one Jack Schwartz. Among his writings was an unpublished Chapter 9 of his compilers book, on parsing natural language, which I of course read.My thesis was on Earley's algorithm applied to natural language. It quickly became apparent that the constraints on phrase structure rules had to be expressed and that one could do that with fairly simple operations on vectors of features, where among the features were what I called the “cores” of the constituents, since they bundled many of the relevant features. My “core” was what linguists came to call “head.” Years later, I ran across Chapter 9 again and reread it, and realized that all the ideas in my thesis were there. So when in 1987 Schwartz told someone that I had anticipated head-driven phrase structure grammar, that was his way of saying he had anticipated head-driven phrase structure grammar.My first job was at Yale University as a very temporary instructor—I think the position is now called “post-doc.” Over the course of the year I became convinced that syntax was a solved problem—something I still believe. But that left me adrift for problems to work on. I became discouraged, and myself again about that Then late one just as I was about to go a graduate Fred came into my to a of a that until that One of the we was a of the of in was before and but similar by the and the English But a of that of the of that could be into a in a In any this a that my research for the next or I at trying to understand how we understand by thesis I what we in the course of an did not what we do not This in in an (Hobbs one from trying to of for the and how problem in the that to at One could say that the of my career a of up and that in of the of and and the specification of I to where I the of and and of at They were to the of of early with on But the problem in a is we trying to an English like is we really to write The is that is a and to or so we left with a with not a of to be to this is to not as a but as an that of have developed for For example, knowing about and and so we might the the wanted to a by the This me because it like we were a with logic we a to or For would we have to It to me that we should by two a could this of by means of if that is the or of we can is by is than It's all applied to where the with over a of to which we go in with can be seen in the could that what information is by the It is a an and a and it the is in by means of the can give this a could call it something like But not it and the the that it that called the who But he at of and He would not have as a I a position because I was and I called that I'm and more I would probably call it something like or or at of I was by the near that many in the early of could be just as we to test our there is probably a out there that But the is we in it. we there is there. no So it doesn't very what we believe. can all sorts of and be of and the we in the of the and we were of of the you to the of and what do you have So should we give any at all to our about what and what not that that can be about in a of since that makes it so to and about the of natural language of this and similar to is that the of a is a of with one there is a The is would be the is not would be But so it would that is not is is that not say that is It that is a of The or not in the and if it that is one of means that we have to the of a and and have But the of is the the of is the of the for on whether or not an really is as the and As a to the the other that in there is an in whether the in the or in the For example, in The did not to the the is the that is is and is that and all of the of a very of In to the of and by a the only is with This us a of The For the The the the and we from the the we that the is a this to that we the the as a we have that and I had the to in a group on and of some of the most of language in the and Fillmore, Kay, and For me the high in and one of the high points in my was when the at the my as (Hobbs and as a to the over the read and on or and other that came to be by linguists the of a would be this I was with the problem of how we the of we as we understand a The that most was that we to to problems of the we were in the But what was there to this of How would you know if list was that the most to me like that a applied to one or more This to three sorts of is the is This the problems of the of like and and the of the in is the is This the problems of and that structure is a of in the In what way the and about and would to be of This the problems of and This of problems I called They problems that the of but they for the the and never on probably because no one saw this of problems as a natural I was about during years was the structure of in that structure out of In this I was very by the work of the linguists and I with the this and we called this of structure and the and and at the University of to be my at and and at were very work the structure of as out of the or of from artificial In work with and work with I to to the of and to and I called this of structure in all three of and was that a was by the of and solve even problems like one had to to a of the where the only for the was that they led to a the of 1987 at SRI we a group on reading the classic by in to for example, and and like as well as work by and at and and at and that to be an similar to Among the in our group were and the who was at the about what we were and and about how to from information in and how to the best of a late one in October 1987 came into my to say that he he had the to all our He his algorithm for It me as the of computational a that not me of it gave us a of what the of a of It gave us a for what to and not The was the most for what would the and an was if and only if it to that my way that I a more In the next few I saw how one would all the and and problems in this In with in the next few it became apparent how one could syntax into the (Hobbs et al. the early 1990s I saw an in a for It a by the a and at a in which the of a is on the and a is in the of another we see we it by up with the best for the was a that the and the into the other someone the and the was into the with a But this is not as an because we have to two and the than just one The first is because it is more this the of the story. is another to be is this in the The is that it is an means there was an in the and they very well could have the and the than for the of a to the for could call the first the It the of the the information by the could call the the It the at that both to be discovered if the is to be that is this The is an trying to prove by that the in a can as well as by to what is the that the of the in that among the is another is to be is the a to a a of the best for an is that it is an at can this with the if is an of a and a the that a some to the of The first in the is the into the of an is the of the The two the into the is the this that the this is that it some or is a a the is in the This is where that It what and I called the in with what is going on in the next of when we the of into using the This that if is a and is a and there is a and the is a a from the we on this we an of by it into the and the just the of that two or a of I have called the and similar to other of that have up the is to the information that can be by the of Structure and a of like and like The first is what is by the is what the is using to do. the by is expressed with in a an or a This not a the two do not is with of a in this a or It out in and this is where syntax in of can a of As we saw at the of is a of rules that us to and complex with the of The best of a is the to us by The best for an is that it is to the of a out in all that to be is the It was the of the as that the best the best of the solved the problems as a I an for that but one should the to The to the three could be an or a a could be an or a and could be a on the or a in a we have with a that have and the most is by there is an and that an we to there is on the to the we to there. that the as a by of the that in the The a the of this a the is so and so it more think there three for were not to from which could for were was a of an of problems in the few now and for of a for into a work by and as a problem in on work by and with this is that when we from a to the we got a of two of there have efforts to and from a number of different to for natural language have had at But efforts to a by language very have to very and and developed a of into and with in into by et al. and she and her have that an using a from is with the best of the in and take on a for is in the next understand so well because we know so one of the problems in the of language is how we our of language and the to This into two How do we the for How do we this in the of had a with Eugene in the early 1990s in which I I the of is a solved The is He with both that the first problem was now the most important of But he that he of that and he had his research I for two I think the of that we at the very of a for can only be by and be by any I think there is that the more than we might at the to the first we to or the as in The on the test from to the we would to be to the someone a on the someone a on the someone a on the In we to and we don't to So we would like to have the same work for the The from to southern from to The from to don't a of and a and an of we for this is as from to if and only if there is a with a is and is that some of is at some of is at and of is at some in the at is for me to we any be to discover rules of this and at the same rules of this of I'm be to discover facts as to be before and have But facts like the of formative in came when I was at Yale in the early and myself in the linguistics Among the that a the most were the like and They were the into to become not and the as in from to into a from at to at They also on the of the at as a for many of the that out of but I think were exactly my they for two they were in what they should have in a by on language they a of so that when they of when and didn't up not they theory was in got on the back for I to it in the when and I on up a which we was to be to prove as a theorem that on a of an can the to because is a of and this a of and in is the we were on with back to the back until or since which the of my It is not research to because in to is very One to that would be by in the next to For example, I was to work with like and on the a of (Hobbs and for on the and on the for my work with and on of the to an of structure called et al. and this led to work with and at on the for an among and my work on or and and work or and My work with on and by over the most by But some of the research you don't about until for out and you work on other My on and (Hobbs and on and of (Hobbs were like is to what I have to call (Hobbs It is not to into It is not to as to in a theory of a theory of of and a theory of and the have to be in so we can not only but also be to with the structure of the is have the to the of the language. have the we have or the in of the in the example, is the to the is a theory of that the and is the that the with the I very and in for the and the is a of other This is to like a like a book, like a and like a It is by a of a of of the a of among and the as a and The at an the to a in a the and different give us different for is at the back of the on a at in an is now at in a The is at the of the of an that at that at a She was at in his at is in this we into a for about the like and of The that into that and should have an in of is a of of and not the same there is an The is in fact, on the is one way to the on the for an and the complex all the and that have to or in for the to say that a the to go on. But many other be in the can't be the to be the to be on in the and so on. The two of a complex that when in the complex or so the and that that is in the complex is relevant in a that can be is a or but in life is because we never can in by is a or It out of a complex a that in a is the part of the the that the in most is the that the to on. the of and but not because in something that or us now a few in of The as in means not not to In most go just means a of as in I go is of a of constraints to do means that it is not the that the constraints not to at means that not to from at from means that there is a from at The is to from a whether a from it or For example, from the in was we would like to be the The the go the of this using the we just with rules from the saying if something it to not and if there is a from a that no and of give of from work I have with on or how we think we have developed to and in and and of the and and I on the position that in a have we to we the we the and if go we our and the two of The first that if an a and that the to have as a The is a similar for the they help can be in of of and can three of the you help someone when you do an that is in a complex for one of In this by as his is like the first with the that the the in the of the For example, if I take a I help him a of but not is no part of his to The third when the and in a as in someone a can the as in to do as the that be and it be a one to one of at is to to do and to have that trying to be to do is to to do and have not in about we to know a about the structure and of the The theory of and us a on because the structure of an a to a For example, the of a is to this by it into two a the and the the by a to the and the in the of a of the most of is our about it to a theory of I think it and of it In we can in of what and what they like more and for a we an of that the and we an of This is the we to when when we that in the relevant especially about what So one of is a of to do more because our is in of is that we not very to a of our should we can be in similar is the of in and and can be seen as to a the of the where a is something that to be with an How it be before we to in the I have I my one of in I the was that the was years we to a system on I the was years So now I myself to be I think a would some of in about

We are developing an ontology of microsocial concepts for use in an instructional system for teaching cross-cultural communication. We report here on that part of the ontology relating to interpersonal relationships. We first explicate the key concepts of commitment, shared plans, and good will. Then in terms of these we present a formal account of simple exchanges, the host-guest relationship, and friendship.

We describe our effort in defining change-of-state words from Core WordNet by constructing axioms anchored in core theories that are crucial in characterizing event words, including change of state, composite entity, scales and event structure. Our methodology consists of three steps: Analyzing the structure of a word's WordNet senses, writing axioms for the most general senses, and testing the axioms on textual entailment pairs. We also describe some common issues we faced and decisions we had to make during axiomatization. We look at two specific textual entailment examples in detail to illustrate the power of this method.

This paper is a study of causation as it occurs in different domains and genres of discourse. There have been various initiatives to extract causality from discourse using causal markers. However, to our knowledge, none of these approaches have displayed similar results when applied to other styles of discourse. In this study we evaluate the nature of causal markers - specifically causatives, between corpora in different domains and genres of discourse and measure the overlap of causal markers using two metrics - Term Similarity and Causal Precision. We find that causal markers, specially causatives (causal verbs) are extremely domain dependent, and moderately genre dependent.

This article presents our work on constructing a corpus of news articles in which events are annotated for estimated bounds on their duration, and automatically learning from this corpus. We describe the annotation guidelines, the event classes we categorized to reduce gross discrepancies in inter-annotator judgments, and our use of normal distributions to model vague and implicit temporal information and to measure inter-annotator agreement for these event duration distributions. We then show that machine learning techniques applied to this data can produce coarse-grained event duration information automatically, considerably outperforming a baseline and approaching human performance. The methods described here should be applicable to other kinds of vague but substantive information in texts.

In the context of developing formal theories of commonsense psychology, or how peole think they think, we have developed a formal theory of goals. In it we explicate and axiomatize, among others, the goal-related notions of trying, success, failure, functionality, intactness, and importance.

We have constructed a corpus of news articles in which events are annotated for estimated bounds on their duration. Here we describe a method for measuring inter-annotator agreement for these event duration distributions. We then show that machine learning techniques applied to this data yield coarse-grained event duration information, considerably outperforming a baseline and approaching human performance.

In this paper, we present our work on creating a complete set of rules for temporal arithmetic mixing months and days based on the "history-dependent intuition". Many examples are presented for demonstrating how the rules are used and how the computations satisfy various desired arithmetic properties, such as the subtraction, commutativity and associativity properties. A notion of "days lost" (DL) is proposed to concisely keep track of the history of the temporal arithmetic computation and explain possible inconsistencies in terms of different desired properties

It has been observed that when people make crude estimates, they feel comfortable choosing between alternatives that differ by a half-order of magnitude (e.g., were there 100, 300, or 1000 people in the crowd?) and less comfortable making a choice on a more detailed scale, with finer granules, or on a coarser scale (like 100 or 1000). In this article, we describe two models of choosing granularity in commonsense estimates, and we show that for both models, in the optimal granularity, the next estimate is three to four times larger than the previous one. Thus, these two optimization results explain the commonsense granularity. © 2006 Wiley Periodicals, Inc. Int J Int Syst 21: 843–855, 2006.

In connection with the DAML project for bringing about the Semantic Web, an ontology of time is being developed for describing the temporal content of Web pages and the temporal properties of Web services. This ontology covers topological properties of instants and intervals, measures of duration, and the meanings of clock and calendar terms.

We provide a general account of parallelism in discourse, and apply it to the special case of resolving possible readings for instances of VP ellipsis. We show how several problematic examples are accounted for in a natural and straightforward fashion. The generality of the approach makes it directly applicable to a variety of other types of ellipsis and reference.

Abduction is inference to the best explanation. In the TACITUS project at SRI we have developed an approach to abductive inference, called “weighted abduction”, that has resulted in a significant simplification of how the problem of interpreting texts is conceptualized. The interpretation of a text is the minimal explanation of why the text would be true. More precisely, to interpret a text, one must prove the logical form of the text from what is already mutually known, allowing for coercions, merging redundancies where possible, and making assumptions where necessary. It is shown how such “local pragmatics” problems as reference resolution, the interpretation of compound nominals, the resolution of syntactic ambiguity and metonymy, and schema recognition can be solved in this manner. Moreover, this approach of “interpretation as abduction” can be combined with the older view of “parsing as deduction” to produce an elegant and thorough integration of syntax, semantics, and pragmatics, one that spans the range of linguistic phenomena from phonology to discourse structure. Finally, we discuss means for making the abduction process efficient, possibilities for extending the approach to other pragmatics phenomena, and the semantics of the weights and costs in the abduction scheme.

In the TACITUS project for using commonsense knowledge in the understanding of texts about mechanical devices and their failures, we have been developing various commonsense theories that are needed to mediate between the way we talk about the behavior of such devices and causal models of their operation. Of central importance in this effort is the axiomatization of what might be called commonsense metaphysics. This includes a number of areas that figure in virtually every domain of discourse, such as granularity, scales, time, space, material, physical objects, shape, causality, functionality, and force. Our effort has been to construct core theories of each of these areas, and then to define, or at least characterize, a large number of lexical items in terms provided by the core theories. In this paper we discuss our methodological principles and describe the key ideas in the various domains we are investigating.

To facilitate work in discourse interpretation, the logical form of English sentences should be both close to English and syntactically simple.In this paper i propose s logical notation which is first-order and uonintensional, sad for which semantic tnmslation can be naively compositional.The key move is to expand what kinds of entities one allows in one's ontology, rather than complicating the logical notation, the logical form of sentences, or the semantic translation process.Three classical problemsopaque adverbials, the distinction between de re and de ditto belief reports, and the problem of identity in intensional contexts -are examined for the dil~cuities they pose for this logical notation, and it is shown that the difficulties can be overcome.The paper closes with s statement about the view of semantics that is presupposed by this appro,-'h.

A sophisticated natural language system requires a large knowledge base. A methodology is described for constructing one in a principled way. Facts are selected for the knowledge base by determining what facts are linguistically presupposed by a text in the domain of interest. The facts are sorted into clusters, and within each cluster they are organized according to their logical dependencies. Finally, the facts are encoded as predicate calculus axioms.

The Problem Consider the sentenceIn most democratic countries most politicians can fool most of the people on almost every issue most of the time.

No abstract available.