File size: 25,575 Bytes
6fa4bc9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 |
{
"paper_id": "W98-0126",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T06:06:08.249490Z"
},
"title": "\"Category Families\" for Categorial Grammars",
"authors": [
{
"first": "Mary",
"middle": [],
"last": "Mcgee",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Manchester",
"location": {
"postCode": "M13 9PL",
"settlement": "Manchester",
"country": "U.K"
}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "Categorial Grammars (CGs; Wood 1993), grounded in algebra (Lambek 1958) and mathematical logic (Ajdukiewicz 1935), have rightly pushed to the limit the use of logically and algebraically justifiable rules for the cornbination and alternation of types in describing natural language. However, when TAG trees are mapped to CG categories, tree-families-linguistically wellmotivated objects-can only be mapped to arbitrary category sets.",
"pdf_parse": {
"paper_id": "W98-0126",
"_pdf_hash": "",
"abstract": [
{
"text": "Categorial Grammars (CGs; Wood 1993), grounded in algebra (Lambek 1958) and mathematical logic (Ajdukiewicz 1935), have rightly pushed to the limit the use of logically and algebraically justifiable rules for the cornbination and alternation of types in describing natural language. However, when TAG trees are mapped to CG categories, tree-families-linguistically wellmotivated objects-can only be mapped to arbitrary category sets.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "To capture predictable category alternations, such as noun / adjective alternations, or valency alternations for verbs, this paper proposes extending a CG with non-algebra-preserving rules, comparable to the \"lexical redundancy rules\" of early lexicalist theory. The theoretical argument is backed by an analysis of the degree of compaction which could be achieved by applying such rules to the CG \"Large Lexicon\" developed at IRCS, UPenn. The reduction which could be achieved both in the number of lexical entries and, more significantly, in the number of categories needed is considerable.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "CGs have always included both binary rules (such as function application and function composition) and unary (type-shifting) rules, and indeed the interactions between these two rule types have been involved in many debates within CG. The unary rules have been restricted to those which preserved algebraic identity: type-raising NP to S/ (S\\NP), for example, does not in itself affect the descriptive power of the grammar. However, it is notorious that words can be highly ambiguous as to category, even in a phrase structure grammar with categories of a fairly coarse grain size (such as \"verb\"), but far more so in a CG. One of the 100 central advances of the lexicalist movement in linguistic description (eg Bresnan (ed) 1982 , Gazdar et al 1985 was the recognition and formalization of patterns in the lexicon such as active / passive alternation. Indeed it is ironic that the most extreme of lexicalist grammars has not adopted such lexical rules. The exact semantics of the rewrite arrow is not at issue here. lt is perhaps best taken as a wellformedness constraint or licensing statement along the iines of GPSG meta-rules: \"if that is iegal, so is this\". Nor are we concerned with implementation details such as whether the rules cause expansion at run-time or compile-time. The claim is that these alternations are facts of natural language, and a linguistic theory must have rules to describe them, as indeed most linguistic theories do.",
"cite_spans": [
{
"start": 713,
"end": 730,
"text": "Bresnan (ed) 1982",
"ref_id": null
},
{
"start": 731,
"end": 750,
"text": ", Gazdar et al 1985",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in theory",
"sec_num": null
},
{
"text": "The UPenn Combinatory CG \"Large Lexicon\" (Doran and Srinivas, forthcoming) was created by automatic translation from the large TAG lexicon developed by the TAG Group at the UPenn Institute for Research in Cognitive Science (XTAG Group 1995) . TAG trees were mapped to CG categories, and the result modified by hand, principally by Christy Doran, B. Srinivas, and Mark Steedman. Some debugging remains to be clone, so these figures are approximate:",
"cite_spans": [
{
"start": 223,
"end": 240,
"text": "(XTAG Group 1995)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "36,950 entries l 7 ,960 words 11 POS values 86 CG categories 120 CG category \"families\" effectively about 110,000 entries (word / category pairs)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "Category f amilies are sets of categories which typically and predictably are assigned together to a word, causing the expansion from 37,000 word entries to 110,000 word / category pairs. In the original TAG lexicon, words are assigned tree families, which are linguistically well-motivated objects (Xia et al, in preparation) . In the translation from TAG trees to CG categories, the motivation is lost, and we are left with seemingly arbitrary category sets. lt is these which can be both motivated and compressed using redundancy rules.",
"cite_spans": [
{
"start": 299,
"end": 326,
"text": "(Xia et al, in preparation)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "Here are some example entries from the lexicon. (The index numbers serve to distinguish atoms within each complex category, and have no other significance. I give the corresponding TAG trees for the first entry only.)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "Verbs: each verb stem has one or two block entries, with some redundancy in passive and gerundive categories: ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "nlDEX: ENTRY: POS: CAT: crease/1 crease V S_O\\NP_O HP _0\\ (NP _1/ll_O) NP_O\\NP_1 ,,,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "101 Canadian/1 Canadian A !IP _0/NP _1 tL0/!1_1 #A_WH- Canadian/2 Canadian A CAT: S_O\\NP_O OlP_O\\HP_1)\\((S_O\\HP_2)/(S_1\\NP_3)) F.S: #A_WH-",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "Since the exact figures for this sort of simple numerical compression are entirely dependent on incidental details of the composition of the original lexicon, it is more significant to look at the size of the set of categories used in the lexicon.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "It is well known that CG categories are more detailed, and therefore more numerous, than the traditional categories of phrase structure grammars ( \"verb\" becomes the set S\\NP, (S\\NP) /NP, ( (S\\NP) /NP) /NP, ... , etc.). It is less commonly observed that a single CG category can correspond to more than one PSG category, where different parts of speech have the same syntactic behaviour. For example,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Redundancy rules in practice",
"sec_num": null
},
{
"text": "The scuffiing and miaowing abated.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Intransitive active",
"sec_num": null
},
{
"text": "The food was accepted.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Transitive bare passive",
"sec_num": null
},
{
"text": "That proposal is absurd.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Predicative adjective",
"sec_num": null
},
{
"text": "Pepper is a tabby cat.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Predicative nominal",
"sec_num": null
},
{
"text": "The president is abroad. I refer to these as the senses of a category, and to a category with more than one sense as ambiguous. A primary sense is basic or irreducible, like the first sense (intransitive active) above. A secondary sense is a derived usage which could be predicted or derived by rule from some other category. Thus S_O\\NP _o (transitive bare passive) is derived from (S_O\\NP _)/NP (transitive active) by a passive rule .which systematically reduces the number of argument NPs to a verb by one. The three predicative senses are derived from basic adjectival, nominal, and prepositional categories by rules which are less neat schematically, but do make the appropriate predictions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Predicative pp",
"sec_num": null
},
{
"text": "(Bear in mind that only the structural syntactic category itself is being considered here. Since TAG trees include part-of-speech information, \"similar\" looking trees are distinguished by the part-of-speech that anchors them. In CG categories, since part-of-speech information is not explicitly encoded, it appears that there are redundancies. However, as we saw above, lexical entries in the CCG Large Lexicon contain a POS field, so during lexical access, given a part-of-speech, there will not be any confusion of this nature.) FUrther, structurally identical categories will often be distinguished at a finer grain-siz.e by having different features. The detailed form of any redundancy rules will have to include these.)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Predicative pp",
"sec_num": null
},
{
"text": "Although the proposed redundancy mies do give a worthwhile reduction in the number of categories needed, the number of senses which can be omitted, and the number of ambiguous categories, are more dramatically reduced. The thirteen ambiguous categories which become unambiguous include the example of S_O\\NP _O given above, which keeps only its primary sense of intransitive verb, Iosing four sec-ondary senses, one passive and three predicative. When one considers that at present the first 15 words in the lexicon with this category are: abate, abdicate, aberrant, abhorrent, abide, abject, able, abnormal, abominable, aboriginal, abort, abortive, above, abrasive, , abdicate, abide, abort, above, abroad, abscond, abstain, abut, accede, accelerate, accept, acclimatize, acconl ",
"cite_spans": [
{
"start": 535,
"end": 666,
"text": "are: abate, abdicate, aberrant, abhorrent, abide, abject, able, abnormal, abominable, aboriginal, abort, abortive, above, abrasive,",
"ref_id": null
},
{
"start": 667,
"end": 779,
"text": ", abdicate, abide, abort, above, abroad, abscond, abstain, abut, accede, accelerate, accept, acclimatize, acconl",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Predicative pp",
"sec_num": null
},
{
"text": "John did not want to get up that early.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Negation on auxiliaries",
"sec_num": null
},
{
"text": "He moved her away.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Exhaustive PPs",
"sec_num": null
},
{
"text": "Redundancy rules will not only compress the explicitly given category set, but expand the set implicitly available. Crossing seven of the basic verb categories (intransitive, intrans + particle, intrans + adjective, transitive, trans + PP comp, trans + VP comp, trans + V comp) with five of the derived forms (active, bare passive, by-passive, gerund, gerund + determiner) should give 29 categories (as intransitives have no passive forms). Of these, only 18 are actually given in the current lexicon, presumably due to accidenta! gaps in the corpus data from which _ its parent TAG lexicon was originally derived.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Exhaustive PPs",
"sec_num": null
},
{
"text": "This proposal will not be popular with the logical purists in the CG community. In language engineering terms, it will be necessary to control the appplicability of redundancy rules and to explore their effect on parsing. What I offer here is some quantified evidence, derived from a realistically large large lexicon intended for serious linguistic description, for the nature and scope of the benefits that a categorial grammar could gain from a systematic formalization of predictable lexical relations through lexical redundancy rules or category families.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": null
}
],
"back_matter": [
{
"text": "This work rides on the shoulders of the people who developed the CCG Large Lexicon: I am deeply grateful to Christy Doran and B. Srinivas, in particular, for their generosity and support. Thanks also to David Bree, Jong C. Park, and Mark Steedman. Much of the credit is theirs; any errors or idiocies are mine.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Die syntaktische Konnexit\u00e4t",
"authors": [
{
"first": "K",
"middle": [],
"last": "Ajdukiewicz",
"suffix": ""
}
],
"year": 1935,
"venue": "Stv.dia Philosophica",
"volume": "1",
"issue": "",
"pages": "1--27",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ajdukiewicz, K. 1935. \"Die syntaktische Kon- nexit\u00e4t\". Stv.dia Philosophica 1:1-27; translated as \"Syntactic connexion\" in McCall (ed) Polish Logic. Oxford.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "The Mental Representation of Grammatical Relations",
"authors": [
{
"first": "J",
"middle": [],
"last": "Bresnan",
"suffix": ""
}
],
"year": 1982,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bresnan, J. ed. 1982. The Mental Representa- tion of Grammatical Relations. MIT Press, Cam- bridge, Mass.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Developing a Wide-Coverage CCG System",
"authors": [
{
"first": "Christy",
"middle": [],
"last": "Doran",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Srinivas",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Forthcoming",
"suffix": ""
}
],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Doran, Christy and B. Srinivas. Forthcoming. \"Developing a Wide-Coverage CCG System\" To appear in a CSLI Volume of the Proceedings of TAG+S, ed. Anne Abeille and Owen Rambow.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Generalized Phrase Structure Grammar",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Generalized Phrase Structure Grammar.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "The Mathematics of Sentence Structure",
"authors": [
{
"first": "Oxford",
"middle": [],
"last": "Blackwell",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Lambek",
"suffix": ""
}
],
"year": 1958,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Blackwell, Oxford. Lambek, J. 1958. \"The Mathematics of Sentence Structure\".",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Marciszewski and van Benthem {eds) Categorial Grammar. John Benjamins",
"authors": [],
"year": 1993,
"venue": "American Mathematical Monthly",
"volume": "65",
"issue": "",
"pages": "154--70",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "American Mathematical Monthly 65:154-70; reprinted in Buszkowski, Mar- ciszewski and van Benthem {eds) Categorial Grammar. John Benjamins, Amsterdam. Wood, M.M. 1993. Categorial Grammars.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Consistent Grammar Development Using Partia!-Trcc Descriptions for Lexicalized 'Ifee-Adjoining Grammar\". The XTAG Group",
"authors": [
{
"first": "Fei",
"middle": [],
"last": "Xia",
"suffix": ""
},
{
"first": "Martha",
"middle": [],
"last": "Palmer",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Vijay-Shanker",
"suffix": ""
},
{
"first": "Joseph",
"middle": [],
"last": "Rosenzweig",
"suffix": ""
}
],
"year": 1995,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Xia, Fei, Martha Palmer, K. Vijay-Shanker, Joseph Rosenzweig. In preparation. \"Consis- tent Grammar Development Using Partia!-Trcc Descriptions for Lexicalized 'Ifee-Adjoining Gram- mar\". The XTAG Group. 1995. \"A Lexi- calized Tree Adjoining Grammar for English\". Technical f\u00fc~port IRCS 95-03, Universit~\u2022 o' Pennsylvania. Updated version available at http://www.cis.upenn.edu/ xtag/tr/tech-report.h-",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"num": null,
"type_str": "figure",
"uris": null,
"text": "CGs could and, I believe, should have such type alternation rules. For example: Nominals: a lexical noun can also serve as a noun phrase, or a noun modifier or noun phrase modifier N =>{NP, N/N, NP/NP} Passives: a lexical verb will also have a passive form taking one fewer nominal complement (S\\NP) /NP => S\\NP ((S\\NP)/NP)/NP=> (S\\NP)/NP etc. Gerundives: a verb (function into S) will also have a gerundive form (function into NP) (S\\NP)/NP => (NP\\NP)/NP ((S\\NP)/NP)/NP=> ((NP\\NP)/NP)/NP etc."
},
"FIGREF1": {
"num": null,
"type_str": "figure",
"uris": null,
"text": "stars above, an Englishman abroad keeps only its prepositional sense and loses the gerundive. The remaining ambiguities are entirely reasonable: for example, (S_O\\NP_O)/(S_l\\NP_l) Adverbs Pepper already was demandiny breakfast. Auxiliary verbs She had prodded John's face seueral times. (S_O\\NP_O)\\(S_l\\NP_l) Adverbs Pepper was demanding breakfast already."
}
}
}
} |