{ "paper_id": "P81-1025", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T09:15:10.967832Z" }, "title": "PERSPECTIVES ON PARSING ISSUES", "authors": [ { "first": "Christopher", "middle": [ "K" ], "last": "Riesbeck", "suffix": "", "affiliation": { "laboratory": "", "institution": "Yale University", "location": {} }, "email": "" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "", "pdf_parse": { "paper_id": "P81-1025", "_pdf_hash": "", "abstract": [], "body_text": [ { "text": "in syntactic knowledge is not the same as believing in parsers! The search for a way to assign 8 syntactic structure to a sentence largely independent of the meaning of that sentence has led to a terrible misdirection of labor. And this effect has been felt on both sides of the fence.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "We find ourselves looking for ways to reduce interaction between syntax and semantics as much as possible.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "How far can we drive a purely syntactic (semantic) analyzer, without sneaking over into the enemy camp? Row well can we disguise syntax (semantics) as semantics (syntax)? How narrow a pipe between the two can we set away with? What a waste of time, when we should be starting with bodies of texts, considering the total language analysis picture, and looking for what kinds of knowledge need to interact to understand those texts.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "If our intent in overextendins our theories was to rest their muscle, then I would have no qualms.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "Pushing a mechanism down a blind alley is an important way to study its weaknesses. But I really can't accept this Popperian view of modern computational linguistics. Mechanisms are not driven beyond their limits to find those limits, but rather to grab territory from the other side. The underlying premise is \"If our mechanism X can sometimes do task A, then there is no need for someone else's mechanism Y.\" Occam's razor is used with murderous intent.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "Furthermore, the debate over whether parsers make sense has drastically reduced interaction between researchers.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "Each side sees the other as avoiding fundamental issues, and so the results from the other side always seem to be beside the point.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "For example, when Mirth Marcus\" explains some grmamatical constraint as syntactic processing constraints, be doesn't answer any of the problems I'm faced with.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "And I'm sure Mitch has no need for frame-based, domain-driven partial language analysis techniques.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "But believing", "sec_num": null }, { "text": "has not arisen because we have been forced to specialize.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "We simply don't know enough to qualify for an information explosion yet. Computational linguistics doesn't have hundreds of journals in dozens of languages.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "It's a young field with only a handful of people working in it.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "Nor is it the case that we don't have things to say to each other.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "But --end here's the rub --some of the most useful things that each of us knows are the things that we don't dare tell.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "By that I mean that each of us knows where our theories fall apart, where ve have to kludge the programs, fudge the inputs, or wince at the outputs.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "That kind of information could be invaluable for suggesting to the others where to focus their attentions.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "Unfortunately, even if we became brave enough to talk about, even emphasize, where we're having problems, the odds are low that we would consider acceptable what someone else proposes as a solution.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This situation", "sec_num": null }, { "text": "Yes, very much so, even if all you are interested in is a good computer program. The reason why was neatly captured in ~rinciDles of Artificia~ lnte~lieence: \"language has evolved as a c~unication medium between intelliaen~ beings\" (Nilsson, p.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "2). That is, natural language usage depends on the fact that certain things can be left ambiguous, left vague, or just left out, because the hearer knows almost as much as the speaker. Natural language has been finely tuned to the co-,-unicative needs of human beings.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "We may have to adapt to the limitations of our ears and our vocal chords, but we have otherwise been the masters of our language. This is true even if there is an innate universal grmmuar (which I don't believe in).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "A universal grammar applies few constraints to our use of ellipsis, ambiguity, anaphora, and all the other aspects of language that make language an efficient means for information transfer, end a pain for the progr----er.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "Because language has been fitted to what we do best, I believe it's improbable that there exist processes very unlike what people use to deal with it.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "Therefore, while I have no intention of trying to model reaction time data points, I do find human behavior important for two kinds of information.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "First, what do people do well, how do they do it, and how does language use depend on it? Second, what do people do poorly, and how does language use get around it?", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "The question '~ow can we know what human processing is really like?\" is a non-issue.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "We don't have to know what human processing is really like. But if people can understand texts that leave out crucial background facts, then our programs have to be able to infer those facts. If people have trouble understanding information phrased in certain ways, then our programs have to phrase it in ways they can understand.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "At some level of description, our programs will have to be \"doing what people do,\" i.e., filling in certain kinds of blanks, leaving out certain kinds of redundancies, and so on. But there is no reason for computational linguists to worry about how deeply their programs correspond to human processes.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "IS SIMULATION OF HUMAN PROCESSING IMPORTANT?", "sec_num": null }, { "text": "People have been predicting (and waiting for) great benefits from parallelism for some time. Personally, I believe that most of the benefits will come in the area of interpretation, where large-scale memory search, such as Scott Fahlman has been worrying about, are involved. And, if anything, improvements in the use of semantics will decrease the attractiveness of syntactic parsing.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WILL PARALLEL PROCESSING CHANGE THINGS?", "sec_num": null }, { "text": "But I also think that there are not that many gains to be had from parallel processing.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WILL PARALLEL PROCESSING CHANGE THINGS?", "sec_num": null }, { "text": "Hash codings, discrimination trees, and so on, already yield reasonably constant speeds for looking up data.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WILL PARALLEL PROCESSING CHANGE THINGS?", "sec_num": null }, { "text": "It is an inconvenience to have to deal with such things, but not an insurmountable obstacle.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WILL PARALLEL PROCESSING CHANGE THINGS?", "sec_num": null }, { "text": "Our real problems at the moment are how to get our systems to make decisions, such as \"Is the question \"How many times has John asked you for money?\" rhetorical or not?\" We are limited not by the number of processors, but by not knowing how to do the job.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WILL PARALLEL PROCESSING CHANGE THINGS?", "sec_num": null }, { "text": "Yes, and adversely.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "To partially contradict my statements in the last paragraph, we've been overly concerned with how to do things with existing hardware and software.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "And we've been too impressed by the success computer science has had with syntax-driven compilation of programming languages.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "I1 is certainly true that work on grammars, parsers, code generators, and so on, have changed compiler generation from maesive multi-man-year endeavors to student course projects. If compiler technology has benefited so much from syntactic parsers, why can't computational linguistics?", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "The problem here is that the technology has not done what people think it has.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "It has allowed us to develop modern, well-structured, task-oriented languages, but it has not given us natural ones. Anyone who has had co teach an introductory progru~ing course knows that. High-level languages, though easier to learn than machine language, are very different from human languages, such as English or Chinese. They show instead what we get if we restrict ourselves to simplistic methods.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "TtI.~E LINGUISTIC PERSPECTIVE HAVE OUR TOOLS AFFECTED US?", "sec_num": null }, { "text": "My working assumption is that the syntactic knowledge used in comprehension is at most context-free and probably a lot less, because of memory limitations. This is mostly a result of semantic heuristics taking over when constructions become too complex for our cognitive chunking capacities. But this is not a critical assumption for me.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO WE PARSE CONTEXT-FREELY?", "sec_num": null }, { "text": "Since I don't believe in the pure gran~atical approach, I have to replace this last set of questions with questions about the relationship between our knowledge (linguistic and otherwise) and the procedures for applying i1.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "; ~rE~AC'~ ONS", "sec_num": null }, { "text": "Fortunately, the questions still make sense after this substitution.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "; ~rE~AC'~ ONS", "sec_num": null }, { "text": "Of course.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO OUR ALGORITHMS AFFECT OUR KNOWLEDGE STRUCTURES?", "sec_num": null }, { "text": "In fact, it is often hard to decide whether some feature of a system is a knowledge structure or a procedural factor. For example, is linear search a result of data structures or procedure designs?", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO OUR ALGORITHMS AFFECT OUR KNOWLEDGE STRUCTURES?", "sec_num": null }, { "text": "CAN WE TEST ALGORITHMS/KNOWLEDGE STRUCTURES SEPARATELY?", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO OUR ALGORITHMS AFFECT OUR KNOWLEDGE STRUCTURES?", "sec_num": null }, { "text": "We do indeed try experiments based on the shape of knowledge structures, independently of bow they are used (but I think that most such experiments have been inconclusive).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO OUR ALGORITHMS AFFECT OUR KNOWLEDGE STRUCTURES?", "sec_num": null }, { "text": "I'm not sure what it would mean, however, for a procedure to be validated independently of the knowledge structures it works with, since until the knowledge structures were right, you couldn't tell if the procedure was doing the right thing or not.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DO OUR ALGORITHMS AFFECT OUR KNOWLEDGE STRUCTURES?", "sec_num": null }, { "text": "If I were trying to deal with this questio n on Erratical grounds, I wouldn't know what it meant. Cr~ars are not processes and hence have no direction. They are abstract characterizations of the set of well-formed strings.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "From certain classes of gra-w-ars 106 one can mechanically build recognizers and rando~ generators.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "But such machines are not the gra-~ars, and a recognizer is manifestly not the same machine as a generator, even though the same grammar may underlie both.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "Suppose ve rephrase the question as '~hy do we have separate knowledge structures for interpretation and production?\"", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "This presupposes that there are separate knowledge structures, and in our current systems this is only partially true.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "Interpreting and production programs abound in ad hoc procedures that share very little in common near the language end.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "The interpreters are full of methods for guessing at meanings, filling in the blanks, predicting likely follow-ups, and so on. The generators are full of methods for eliminating contextual items, picking appropriate descriptors, choosing pronouns, and so on. Each has a very different set of problems to deal with.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "On the other hand, our interpreters and generators do share what we think is the important stuff, the world knowledge, without which all the other processing wouldn't be worth a partridge in a parse tree. This viii have account for how we can interpret words without being ready to use them.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "For example, from a sentence like \"The car swerved off the road and struck a bridge abutment,\" we can infer that an abutment is a noun describing some kind of outdoor physical object, attachable to a bridge. This would be enough for interpretation, but obviously the generator will need co know more about what an abutment is before it could confidently say \"Oh, look at the cute abutment!\"", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "A final point on sharing.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "There are two standard arguments for sharing at least gr=mmatical information. One is to save space, and the other is to maintain consistency.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "Without claiming that sharing doesn't occur, I would like to point out that both arguments are very weak.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "First, there is really not a lot of grammatical knowledge, compared against all the other knowledge we have about the world, so not that much space would be saved if sharing occurred.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "Second, if the generator derives it's linguistic knowledge from the parser's data base, then we'll have as much consistency as we could measure in people anyway.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "WHY DO WE SEPARATE RECOGNITION AND PRODUCTION?", "sec_num": null }, { "text": "Nilsson, H. (1980) . Princinle~ of Artificia~ Intellisence.", "cite_spans": [ { "start": 12, "end": 18, "text": "(1980)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "REFERENCE", "sec_num": null }, { "text": "Tioga Publishing Co, Palo Alto, California.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "REFERENCE", "sec_num": null } ], "back_matter": [], "bib_entries": {}, "ref_entries": {} } }