File size: 60,076 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 |
{
"paper_id": "W07-0402",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T04:39:17.334072Z"
},
"title": "Extraction Phenomena in Synchronous TAG Syntax and Semantics",
"authors": [
{
"first": "Rebecca",
"middle": [],
"last": "Nesson",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Harvard University Cambridge",
"location": {
"postCode": "02138",
"region": "MA"
}
},
"email": "[email protected]"
},
{
"first": "Stuart",
"middle": [
"M"
],
"last": "Shieber",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Harvard University Cambridge",
"location": {
"postCode": "02138",
"region": "MA"
}
},
"email": "[email protected]"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "We present a proposal for the structure of noun phrases in Synchronous Tree-Adjoining Grammar (STAG) syntax and semantics that permits an elegant and uniform analysis of a variety of phenomena, including quantifier scope and extraction phenomena such as wh-questions with both moved and in-place wh-words, pied-piping, stranding of prepositions, and topicalization. The tight coupling between syntax and semantics enforced by the STAG helps to illuminate the critical relationships and filter out analyses that may be appealing for either syntax or semantics alone but do not allow for a meaningful relationship between them.",
"pdf_parse": {
"paper_id": "W07-0402",
"_pdf_hash": "",
"abstract": [
{
"text": "We present a proposal for the structure of noun phrases in Synchronous Tree-Adjoining Grammar (STAG) syntax and semantics that permits an elegant and uniform analysis of a variety of phenomena, including quantifier scope and extraction phenomena such as wh-questions with both moved and in-place wh-words, pied-piping, stranding of prepositions, and topicalization. The tight coupling between syntax and semantics enforced by the STAG helps to illuminate the critical relationships and filter out analyses that may be appealing for either syntax or semantics alone but do not allow for a meaningful relationship between them.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "1 Introduction Nesson and Shieber (2006) showed how a nowstandard variant of the tree-adjoining grammar (TAG) formalism (multi-component, multiple adjunction, finite-feature-based TAG), when synchronized, leads to a natural analysis of the syntaxsemantics relation, including handling of syntactic movement phenomena such as wh questions and relativization, semantic \"movement\" phenomena such as quantification, quantifier scope ambiguity, and even their interactions as found in pied-piped relative clauses. 1 phenomena were previously viewed as problematic for TAG analyses, leading to the hypothesizing of various extensions to the TAG formalism (Kallmeyer and Romero, 2004 , and work cited therein). Independently, Han (2006a) developed a similar synchronous TAG analysis of piedpiping, providing evidence for the naturalness of the analysis.",
"cite_spans": [
{
"start": 15,
"end": 40,
"text": "Nesson and Shieber (2006)",
"ref_id": "BIBREF6"
},
{
"start": 649,
"end": 676,
"text": "(Kallmeyer and Romero, 2004",
"ref_id": "BIBREF3"
},
{
"start": 719,
"end": 730,
"text": "Han (2006a)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Here, we update the analyses of noun phrases found in the previous works in one simple way, again with no additional formal TAG innovations, and show that it allows a further coverage of extraction and quantification phenomena as well as in-situ wh-phrases and topicalization. We emphasize that no novel formal devices are postulated to achieve this increased coverage -just a simple, natural and uniform change to the canonical structure of NPs and their semantics.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "A word may be useful on the pertinence of this work in a workshop on \"syntax and structure in machine translation\", above and beyond the intrinsic importance of exploring the \"applications of [synchronous/transduction grammars] to related areas including. . . formal semantics\" underlying the workshop. Tree-structured mappings are advocated for machine translation systems because they allow for the expression of generalizations about relationships between languages more accurately and effectively. Evidence for this benefit ought to be found in the ability of the formalisms to characterize the primitive linguistic relationships as well, in particular, the form-meaning relationship for a natural language. The present work is part of a general program to explore the suitability of synchronous grammars for expressing this primary linguistic relationship. Inso-far as it is successful, it lends credence to the use of these formal tools for a variety of language processing tasks, including MT. Insofar as it reveals insufficiencies in the formalism, it may lead to insights in the design or deployment of alternative systems.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "We present a proposal for the structure of noun phrases in Synchronous Tree-Adjoining Grammar (STAG) syntax and semantics that permits an elegant and uniform analysis of a variety of phenomena, including quantifier scope and extraction phenomena such as wh-questions with both moved and in-situ wh-words, pied-piping, stranding of prepositions, and topicalization. Furthermore, the tight coupling between syntax and semantics enforced by grammar synchronization helps to illuminate the critical relationships and filter out analyses that may be appealing for either syntax or semantics alone but do not allow for a meaningful relationship between them.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "We begin in Section 2 with a brief review of synchronous TAG and its application to English syntax and semantics. In Section 3, we present an analysis of quantifier scope that elucidates the relationship between the syntactic and semantic structures and explains an anomaly of previously proposed analyses. We apply the underlying idea from Section 3 to wh-questions in Section 4, showing that an alteration of the standard TAG syntax analysis of whquestions produces the same derived trees while also elegantly modeling in-place wh-words. In Section 5 we present a challenging case for STAG syntax and semantics, the stranding of prepositions. This case is particularly difficult because the syntactic analyses suggested by previous work in STAG syntax do not encapsulate the relationships that appear necessary for the semantics. Our proposed analysis falls out naturally from the revision to the syntax of whwords and respects both Frank's Condition on Elementary Tree Minimality (CETM) and the semantic relationships in the construction. In Section 6 we give an analysis of topicalization that also follows from the underlying ideas of the earlier analyses. We summarize the main ideas of the analysis in Section 7.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "A tree-adjoining grammar (TAG) consists of a set of elementary tree structures of arbitrary depth, which are combined with two operations, substitution and adjunction. Internal nodes in the elementary trees are labeled with a nonterminal symbol. Frontier nodes may be labeled with either terminal symbols or nonterminal symbols annotated with one of the diacritics \u2193 or * . The \u2193 diacritic marks a frontier nonterminal node as a substitution node, the target of the substitution operation. The substitution operation occurs when an elementary tree rooted in a nonterminal symbol A replaces a substitution node with the same nonterminal symbol.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "S N P \u2193 V P N P \u2193 V likes N P John S V P N P \u2193 V likes N P John V P Adv V P * S N P \u2193 V P N P \u2193 V likes apparently V P Adv S N P \u2193 V P N P \u2193 V likes apparently =\u21d2 =\u21d2",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "Auxiliary trees are elementary trees in which the root and a frontier node, called the foot node and distinguished by the diacritic * , are labeled with the same nonterminal A. The adjunction operation involves splicing an auxiliary tree in at an internal node in an elementary tree also labeled with nonterminal A. Trees without a foot node, intended for substitution rather than adjunction into other trees, are called initial trees. Examples of the substitution and adjunction operations on sample elementary trees are shown in Figure 1 . For further information, refer to Joshi and Schabes (1997) .",
"cite_spans": [
{
"start": 586,
"end": 600,
"text": "Schabes (1997)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [
{
"start": 531,
"end": 539,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "Synchronous TAG (Shieber, 1994; Shieber and Schabes, 1990) extends TAG by taking the elementary structures to be pairs of TAG trees with links between particular nodes in those trees. Derivation proceeds as in TAG except that all operations must be paired. That is, a tree can only be substituted or adjoined at a node if its pair is simultaneously substituted or adjoined at a linked node. We notate the links by using boxed indices i marking linked nodes. Figure 2 : An English syntax/semantics STAG fragment (a), derived tree pair (b), and derivation tree (c) for the sentence \"John apparently likes Mary.\"",
"cite_spans": [
{
"start": 16,
"end": 31,
"text": "(Shieber, 1994;",
"ref_id": "BIBREF9"
},
{
"start": 32,
"end": 58,
"text": "Shieber and Schabes, 1990)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [
{
"start": 458,
"end": 466,
"text": "Figure 2",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "Adv V P * t t * S N P \u2193 V P N P \u2193 V t e \u2193",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "As first described by Shieber and Schabes (1990) , STAG can be used to provide a semantics for a TAG syntactic analysis by taking the tree pairs to represent a syntactic analysis synchronized with a semantic analysis.",
"cite_spans": [
{
"start": 34,
"end": 48,
"text": "Schabes (1990)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "For example, Figure 2 (a) contains a sample English syntax/semantics grammar fragment that can be used to analyze the sentence \"John apparently likes Mary\". The node labels we use in the semantics correspond to the semantic types of the phrases they dominate.",
"cite_spans": [],
"ref_spans": [
{
"start": 13,
"end": 21,
"text": "Figure 2",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "Figure 2(c) shows the derivation tree for the sentence. Substitutions are notated with a solid line and adjunctions are notated with a dashed line. Each link in the derivation tree specifies a link number in the elementary tree pair, providing the location at which the operations take place. In this case, the tree pairs for the noun phrases John and Mary substitute into the likes tree pair at links 3 and 4 , respectively. The word apparently adjoins at link 2 . The tree pair so derived is shown in Figure 2 (b). The resulting semantic representation can be read off the right-hand derived tree by treating the leftmost child of a node as a functor and its siblings as its arguments. Our sample sentence thus results in the semantic representation apparently(likes( john, mary)).",
"cite_spans": [],
"ref_spans": [
{
"start": 503,
"end": 511,
"text": "Figure 2",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Introduction to Synchronous TAG",
"sec_num": "2"
},
{
"text": "We start by reviewing the prior approach to quantifier semantics in synchronous TAG. Consider the sentence \"Everyone likes someone.\" We would like to allow both the reading where some takes scope over every and the reading where every takes scope over some. We start with the proposal of Shieber and Schabes (1990) , which used multi-component TAG for the semantic portion of a synchronous TAG. Each quantified noun phrase has a two-component tree set as its semantics. One component introduces the variable quantified over in the scope of the quantifier; the other adjoins over the scope to provide the quantifier and restriction. Williford (1993) explored the use of multiple adjunction (Schabes and Shieber, 1993) to achieve scope ambiguity. Since the scope components of subject and object noun phrases adjoin at the same location in the semantic tree, they give rise to a systematic ambiguity as to which dominates the other in the derived tree, reflecting the semantic scope ambiguity of the sentence; the derivation tree itself is therefore a scope neutral representation. Previous work by Han (2006a; 2006b) and Nesson and Shieber (2006) describe this approach in detail, showing its applicability to a range of semantic phenomena.",
"cite_spans": [
{
"start": 300,
"end": 314,
"text": "Schabes (1990)",
"ref_id": "BIBREF8"
},
{
"start": 632,
"end": 648,
"text": "Williford (1993)",
"ref_id": "BIBREF10"
},
{
"start": 689,
"end": 716,
"text": "(Schabes and Shieber, 1993)",
"ref_id": "BIBREF7"
},
{
"start": 1097,
"end": 1108,
"text": "Han (2006a;",
"ref_id": "BIBREF0"
},
{
"start": 1109,
"end": 1115,
"text": "2006b)",
"ref_id": "BIBREF1"
},
{
"start": 1120,
"end": 1145,
"text": "Nesson and Shieber (2006)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "A range of research has proceeded in an alternative line of using complex-feature-based TAGrather than synchronous TAG -for TAG semantics (Kallmeyer and Romero, 2004 , and work cited therein). Semantic representations are carried in features associated with nodes. Nonetheless, multicomponent TAG with separate trees for bound position and scope is used here too. However, the two trees are syntactic trees, the quantified NP tree and a vestigial S tree, respectively. (An example is shown in Figure 6 .) In such analyses, the single-node auxiliary S tree is used for the scope part of the syntax in order to get the desired relationship between the quantifier and the quantified expression in features threaded through the derivation tree and hence in the semantics.",
"cite_spans": [
{
"start": 138,
"end": 165,
"text": "(Kallmeyer and Romero, 2004",
"ref_id": "BIBREF3"
}
],
"ref_spans": [
{
"start": 493,
"end": 501,
"text": "Figure 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "The present analysis marries these two ap- , and derived trees (c) for the sentence \"Everyone likes someone\". Note that the derivation tree is a scope neutral representation: depending on whether every or some adjoins higher, we obtain different semantic derived trees and scope orderings.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "proaches. Like the previous STAG work, we propose a solution in which a multi-component tree set provides semantics for quantified phrases, with multiple adjunction providing scope ambiguity. Like the complex-feature-based approach, we reflect the multi-component structure in the syntax as well. It is this single change in the analysis that makes possible the coverage of the wide range of phenomena we describe here. Combining these two approaches, we give both the syntactic and semantic trees for quantifiers two parts, as depicted in Figure 3(a) . In the semantics, the top part corresponds to the scope of the quantifier and attaches where the quantifier takes scope. The bottom part corresponds to the bound variable of the quantifier. By multiply adjoining the scope parts of the semantic trees of the quantifiers at the same location in the likes tree, we generate both available scope readings of the sentence. 2 Correspondingly on the syntax side, an NP tree provides the content of the noun phrase with a vestigial S tree available as well. Prior to the analyses given in this paper, the use of two trees in the quantifier syntax was an arbitrary stipulation used to make the semantic analysis possible. The pairing of the upper tree in the syntax with the scope tree in the semantics explicitly demonstrates their relationship and leads naturally to the exploration of non-degenerate upper trees in the syntax that we explore in this paper.",
"cite_spans": [],
"ref_spans": [
{
"start": 540,
"end": 551,
"text": "Figure 3(a)",
"ref_id": "FIGREF3"
}
],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "In order to use these multi-component quantifiers, we change the links in the elementary trees for verbs to allow a single link to indicate two positions in the syntax and semantics where a tree pair can adjoin, as shown in Figure 3(a) . We add four-way links and drop the two-way links used by the unquantified noun phrases in the first example. This choice forces all noun phrase tree pairs to be multicomponent in the syntax and semantics. Essentially, all noun phrases are \"lifted\"\u00e0 la Montague. We explore the consequences of this in Section 6.",
"cite_spans": [],
"ref_spans": [
{
"start": 224,
"end": 235,
"text": "Figure 3(a)",
"ref_id": "FIGREF3"
}
],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "We turn now to the ramifications of this new syntactico-semantic STAG representation, showing its utility for a range of phenomena.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Quantifier Scope",
"sec_num": "3"
},
{
"text": "The structure we propose for quantifiers suggests a new possibility for the TAG analysis of wh-words. We propose to simply treat wh-words as regular noun phrases by making them a multi-component tree set with an auxiliary tree that adjoins at the root of the verb tree and contains the lexical content and an initial tree with an empty frontier that substitutes at the argument position. This syntactic tree set can",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "S * W H N P x e t who t * x S who S * W H who N P S or which N P S W H N \u2193 S * N P S * S N P which N P W H N \u2193 t which x t t * e",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "x e x e, t \u2193 or Figure 4 : Elementary tree pairs for who and which. The left and middle tree sets are the syntactic alternatives used to model wh-movement and in-situ wh-words. The tree sets on the right provide the semantics.",
"cite_spans": [],
"ref_spans": [
{
"start": 16,
"end": 24,
"text": "Figure 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "2 3 4 S N P \u2193 V P N P \u2193 V likes 3 1 4 2 3 4 S V P N P \u2193 V likes 3 1 4 S W H \u2193 2 4 S N P \u2193 V P V likes 3 1 4 S W H \u2193 N P N P",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "Figure 5: Traditional elementary trees for the verb likes. Using a revised, elementary syntax tree set for wh-words like who, only the left tree is necessary.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "be paired with a multi-component semantic tree set that has an auxiliary tree containing the scope part and an initial tree that contains the bound variable. Wh-questions with the wh-word in place can be elegantly modeled with an alternative syntactic tree set in which the auxiliary tree has no lexical content and the wh-word is on the frontier of the initial tree that substitutes into the argument position. The semantic tree sets for both syntactic variations is the same. These trees are shown in Figure 4 . Besides the incorporation of a semantics, the basic analyses for wh-questions familiar from TAG syntax are otherwise unchanged because the top piece of the syntax tree set still ends up at the root of the main verb in sentences such as the following:",
"cite_spans": [],
"ref_spans": [
{
"start": 503,
"end": 511,
"text": "Figure 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "(1) Who likes Mary?",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "who(x, likes(mary, x)) Note that in Sentence 3 thinks is not constrained to appear to the right of who in the syntax, because thinks and who both adjoin at the same location in the syntax. However, we can use a feature to force embedding verbs to adjoin lower than wh-words. The same situation exists in Sentence 4, though only in the semantics; the order of words in the syntax is well-defined but the multiple adjunction of the scope of who and the scope of each underspecifies the scope ordering between them. Both scope orderings are indeed arguably valid. Again, the preferences for certain orderings can be regulated using a feature. These issues highlight the many open questions about how to combine quantification and wh-terms, but also provides a first step towards their analysis within a concise STAG construction.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "Our approach has several distinct advantages. First, it allows wh-words to be analyzed in a way that is uniform with the analysis of other noun phrases and allows us to simplify the lexical entries for verbs. In the traditional TAG analysis, wh-words substitute into specialized lexical trees for verbs that add an additional frontier node for the wh-word and abstract over one of the arguments of the verb by adding an empty terminal node at the frontier. Our revision to the elementary trees for wh-words allows us to remove several tree pairs from the elementary tree sets for verbs such as like. Instead of requiring an elementary tree pair for declarative sentences and an additional elementary tree for each argument",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "S' * S' WH whom NP John \u2193 S does VP V like S * Det a N * NP * N PP picture of \uf8f1 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f3 a b c d e \uf8fc \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fd \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fe \uf8f1 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f3 a b c d e \uf8fc \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fd \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fe \uf8f1 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f3 a b c d e f \uf8fc \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fd \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fe \uf8f1 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f3 a b c d e f \uf8fc \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fd \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8f4 \uf8fe WH\u2193 S NP NP NP [ ] NP",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "Figure 6: Kallmeyer and Scheffler's syntactic analysis for Sentence 6.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "that can be replaced by a fronted wh-word to form a question (as shown in Figure 5 ), we can use just the single declarative sentence elementary tree. Second, it provides a simple and elegant characterization of the syntax and semantics of whmovement and the relationship between fronted and in-place wh-words. Using the alternative syntax tree set given in Figure 4 we model in-place use of whwords as in Sentence 5 while still maintaining the usual semantic analysis:",
"cite_spans": [],
"ref_spans": [
{
"start": 74,
"end": 82,
"text": "Figure 5",
"ref_id": null
},
{
"start": 358,
"end": 366,
"text": "Figure 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "(5) John likes who?",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "who(x, likes(x, john))",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Wh-questions",
"sec_num": "4"
},
{
"text": "Sentence 6 presents a particularly challenging case for TAG semantics. The problem arises because who must contribute its bound variable, x, to the noun phrase \"a picture of x\". However, in the standard syntactic analysis who substitutes into the likes tree, and in any reasonable semantic analysis, who takes scope at the root of the likes tree.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "(6) Who does John like a picture of? who(x, a(y, and(picture(y), of (x, y)), likes( john, y))) Kallmeyer and Scheffler (2004) propose a syntactic analysis in which \"a picture of\" adjoins into the syntactic tree for \"likes\". The syntax for this analysis is shown for comparison in Figure 6 .",
"cite_spans": [
{
"start": 95,
"end": 125,
"text": "Kallmeyer and Scheffler (2004)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [
{
"start": 280,
"end": 288,
"text": "Figure 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "Associated with the syntactic analysis is a semantic analysis, which differs from ours in that all of the semantic computation is accomplished by use of a flexible set of features that are associated with nodes in the syntactic trees. This analysis maintains Frank's Constraint on Elementary Tree Minimality (CETM) if one analyzes the prepositional phrase as a complement of picture but it does so at the expense of a straightforward compositional semantics. 4 The source of the problem is that who contributes its bound variable to likes to form an intermediate semantics who (x, likes( john, x) ), then a picture of combines non-compositionally to form the complete semantics given in Sentence 6. Kroch (1989) describes the intuition eschewing this analysis: \"The problem is that under such a derivation, the preposed wh-phrase changes its thematic role with each adjunction and the interpretation of the derived tree is not a simple function of the interpretations of its component elementary trees.\" When we consider the semantics of the two sentences, the anomaly of this analysis becomes apparent. In the first sentence the entity liked by John is referred to by the variable contributed by who. In the second sentence John likes an entirely different entity: the entity referred to by the variable contributed by a. Kallmeyer and Scheffler obtain the correct semantics by making use of non-local TAG operations to have the scope part of a adjoin into likes to capture the semantics of the likes proposition and employing a feature-based mechanism for swapping the variables as necessary.",
"cite_spans": [
{
"start": 459,
"end": 460,
"text": "4",
"ref_id": null
},
{
"start": 699,
"end": 711,
"text": "Kroch (1989)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [
{
"start": 577,
"end": 596,
"text": "(x, likes( john, x)",
"ref_id": null
}
],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "Our revision to the syntax of wh-words provides an alternative way of maintaining the CETM that offers a much simpler semantic analysis. The details of the analysis are given in Figure 7 . We adjoin who into the preposition of at link 1 where it contributes both variable and scope. The tree pair for of attaches to a at link 1 , thus allowing the scope parts of the quantifier a and the wh-word who to end up taking scope over the main verb as in the analysis of prepositional phrases given by Nesson and Shieber (2006) . It also places all the bound variables in the correct propositions without use of non-local operations or additional manipulation. A diagram of the derived syntax and semantics is given in Figure 8 . Figure 7 : The elementary tree pairs and derivation tree for the sentence \"Who does John like a picture of?\". ",
"cite_spans": [
{
"start": 495,
"end": 520,
"text": "Nesson and Shieber (2006)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [
{
"start": 178,
"end": 186,
"text": "Figure 7",
"ref_id": null
},
{
"start": 712,
"end": 720,
"text": "Figure 8",
"ref_id": "FIGREF6"
},
{
"start": 723,
"end": 731,
"text": "Figure 7",
"ref_id": null
}
],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "N P \u2193 V P V",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "N P \u2193 V P V likes N P \u2193 S 3 4 3 1 4 2 Det a N P N \u2193 S * 1 1 2 N P P P P of S * N P * N P \u2193",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Stranded Prepositions",
"sec_num": "5"
},
{
"text": "The insight that allows us to model in-place whwords extends to an elegant analysis of topicalization as well. The vestigial S * tree that we added to the tree set for the syntax of every noun phrase need not always be contentless. Just as we moved the wh-word who from the top tree in its set to the bottom tree to model in-situ wh-words, we can move the lexical content of noun phrases to the top tree in their sets to model topicalization. For instance, the alternative tree pair for Mary shown in Figure 9 provides for an analysis of the sentence (7) Mary, John likes. likes(mary,john)",
"cite_spans": [],
"ref_spans": [
{
"start": 501,
"end": 509,
"text": "Figure 9",
"ref_id": "FIGREF7"
}
],
"eq_spans": [],
"section": "Topicalization",
"sec_num": "6"
},
{
"text": "The analysis interacts properly with that for preposition stranding, so that the sentence follows from the tree pair for a in the same figure.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Topicalization",
"sec_num": "6"
},
{
"text": "In this paper we have proposed a uniform change to the structure of noun phrases in the STAG syntactico-semantic grammar. The formal tools we avail ourselves of comprise synchronous TAG with set-local multicomponent adjunction and multiple adjunction. Nothing more is required.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "7"
},
{
"text": "All noun phrases now have a uniform multicomponent structure in both the syntax and the semantics. In the semantics the top part corresponds to the scope-giving piece provided by the noun phrase and the bottom part to the bound variable or simple noun-phrase meaning. In the syntax, the top part corresponds to the lexical material that should appear moved to the edge of the sentence or clause; the bottom part corresponds to the lexical material that will fill an argument position of some head. By moving lexical material among the pieces of the multi-component set in the syntax, we can simply model phenomena like in-place wh-words and topicalization.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "7"
},
{
"text": "Making the top parts of wh-word tree sets into auxiliary trees allows them to adjoin not just to the main verb but also to heads of modifying clauses, such as prepositional phrases. This allows us to handle more complex sentences like Sentence 6 without violating either the CETM or going beyond simple compositional semantics. In order to allow the scope-giving part of the wh-word to percolate up to the root of the semantics of the main verb, each tree set that it adjoins into on its way must also have a scope part in the semantics to which it can adjoin. Scope carriers, such as prepositions, are therefore also multi-component in the semantics with a top node to which scope-givers can adjoin. One nice property of this analysis is that it predicts the observed facts about disallowed scope orderings in sentences that have three quantifiers, one of which is in a modifying clause. The scope part of the quantifier of the modified clause and the scope part of the quantifier of the modifying clause form an indivisible set as the derivation proceeds so that when they adjoin multiply with the scope part of the unmodified clause, that quantifier cannot intervene between them.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "7"
},
{
"text": "Our synchronous grammar treatment of the syntax-semantic relation with TAG is at least as simple and arguably more accurate than previous TAG proposals, offering treatments of such phenomena as in-situ wh-words, stranded prepositions, and topicalization.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "7"
},
{
"text": "This work was supported in part by grant IIS-0329089 from the National Science Foundation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Nesson and Shieber (2006) provide a more in-depth explanation of the multiple-adjunction-driven approach to scope neutrality in STAG.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The presence of do-support in wh-questions can be handled independently using a feature on the NP node into which the bottom part of the wh-word tree pair substitutes that governs whether and where a do tree adjoins.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "In addition to suggesting a non-compositional semantics, their syntactic analysis makes use of non-local multicomponent TAG in order to achieve the necessary semantic relationships. Although their use of non-local TAG may be benign in terms of complexity, our analysis is set-local. Our proposal therefore simplifies the syntactic analysis while also bringing it in line with a straightforward, compositional semantics.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Pied-piping in relative clauses: Syntax and compositional semantics based on synchronous tree adjoining grammar",
"authors": [
{
"first": "Chung-Hye",
"middle": [],
"last": "Han",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the 8th International Workshop on Tree Adjoining Grammars and Related Formalisms (TAG+ 8)",
"volume": "",
"issue": "",
"pages": "41--48",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chung-Hye Han. 2006a. Pied-piping in relative clauses: Syntax and compositional semantics based on syn- chronous tree adjoining grammar. In Proceedings of the 8th International Workshop on Tree Adjoining Grammars and Related Formalisms (TAG+ 8), pages 41-48, Sydney, Australia.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "A tree adjoining grammar analysis of the syntax and semantics of it-clefts",
"authors": [
{
"first": "Chung-Hye",
"middle": [],
"last": "Han",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the 8th International Workshop on Tree Adjoining Grammars and Related Formalisms (TAG+ 8)",
"volume": "",
"issue": "",
"pages": "33--40",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chung-Hye Han. 2006b. A tree adjoining grammar anal- ysis of the syntax and semantics of it-clefts. In Pro- ceedings of the 8th International Workshop on Tree Adjoining Grammars and Related Formalisms (TAG+ 8), pages 33-40, Sydney, Australia.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Treeadjoining grammars",
"authors": [
{
"first": "K",
"middle": [],
"last": "Aravind",
"suffix": ""
},
{
"first": "Yves",
"middle": [],
"last": "Joshi",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Schabes",
"suffix": ""
}
],
"year": 1997,
"venue": "Handbook of Formal Languages",
"volume": "",
"issue": "",
"pages": "69--124",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Aravind K. Joshi and Yves Schabes. 1997. Tree- adjoining grammars. In G. Rozenberg and A. Salo- maa, editors, Handbook of Formal Languages, pages 69-124. Springer.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "LTAG semantics with semantic unification",
"authors": [
{
"first": "Laura",
"middle": [],
"last": "Kallmeyer",
"suffix": ""
},
{
"first": "Maribel",
"middle": [],
"last": "Romero",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of TAG+7",
"volume": "",
"issue": "",
"pages": "155--162",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Laura Kallmeyer and Maribel Romero. 2004. LTAG se- mantics with semantic unification. In Proceedings of TAG+7, pages 155-162, Vancouver, May.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "LTAG analysis for pied-piping and stranding of wh-phrases",
"authors": [
{
"first": "Laura",
"middle": [],
"last": "Kallmeyer",
"suffix": ""
},
{
"first": "Tatjana",
"middle": [],
"last": "Scheffler",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of TAG+7",
"volume": "",
"issue": "",
"pages": "32--39",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Laura Kallmeyer and Tatjana Scheffler. 2004. LTAG analysis for pied-piping and stranding of wh-phrases. In Proceedings of TAG+7, pages 32-39, Vancouver, May.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Asymmetries in long distance extraction in a tree adjoining grammar",
"authors": [
{
"first": "Anthony",
"middle": [],
"last": "Kroch",
"suffix": ""
}
],
"year": 1989,
"venue": "Alternative Conceptions of Phrase Structure",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anthony Kroch. 1989. Asymmetries in long distance extraction in a tree adjoining grammar. In Mark Baltin and Anthony Kroch, editors, Alternative Conceptions of Phrase Structure. University of Chicago Press.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Simpler TAG semantics through synchronization",
"authors": [
{
"first": "Rebecca",
"middle": [],
"last": "Nesson",
"suffix": ""
},
{
"first": "Stuart",
"middle": [
"M"
],
"last": "Shieber",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the 11th Conference on Formal Grammar",
"volume": "",
"issue": "",
"pages": "29--30",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rebecca Nesson and Stuart M. Shieber. 2006. Sim- pler TAG semantics through synchronization. In Pro- ceedings of the 11th Conference on Formal Grammar, Malaga, Spain, 29-30 July.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "An alternative conception of tree-adjoining derivation",
"authors": [
{
"first": "Yves",
"middle": [],
"last": "Schabes",
"suffix": ""
},
{
"first": "Stuart",
"middle": [
"M"
],
"last": "Shieber",
"suffix": ""
}
],
"year": 1993,
"venue": "Computational Linguistics",
"volume": "20",
"issue": "1",
"pages": "91--124",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yves Schabes and Stuart M. Shieber. 1993. An alterna- tive conception of tree-adjoining derivation. Compu- tational Linguistics, 20(1):91-124.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Synchronous tree-adjoining grammars",
"authors": [
{
"first": "M",
"middle": [],
"last": "Stuart",
"suffix": ""
},
{
"first": "Yves",
"middle": [],
"last": "Shieber",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Schabes",
"suffix": ""
}
],
"year": 1990,
"venue": "Proceedings of the 13th International Conference on Computational Linguistics",
"volume": "3",
"issue": "",
"pages": "253--258",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stuart M. Shieber and Yves Schabes. 1990. Synchronous tree-adjoining grammars. In Proceedings of the 13th International Conference on Computational Linguis- tics, volume 3, pages 253-258, Helsinki.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Restricting the weak-generative capacity of synchronous tree-adjoining grammars",
"authors": [
{
"first": "M",
"middle": [],
"last": "Stuart",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Shieber",
"suffix": ""
}
],
"year": 1994,
"venue": "Computational Intelligence",
"volume": "10",
"issue": "4",
"pages": "371--385",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stuart M. Shieber. 1994. Restricting the weak-generative capacity of synchronous tree-adjoining grammars. Computational Intelligence, 10(4):371-385, Novem- ber.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Application of synchronous treeadjoining grammar to quantifier scoping phenomena in English",
"authors": [
{
"first": "Sean",
"middle": [],
"last": "Williford",
"suffix": ""
}
],
"year": 1993,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sean Williford. 1993. Application of synchronous tree- adjoining grammar to quantifier scoping phenomena in English. Undergraduate Thesis, Harvard College.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"num": null,
"type_str": "figure",
"text": "Example TAG substitution and adjunction.",
"uris": null
},
"FIGREF3": {
"num": null,
"type_str": "figure",
"text": "The elementary tree pairs (a), derivation tree (b)",
"uris": null
},
"FIGREF4": {
"num": null,
"type_str": "figure",
"text": "Which person does John like? 3 which(x, person(x), likes(x, john)) (3) Which person does Bill think John likes? which(x, person(x),thinks(bill, likes(x, john))) (4) Who does each person like? who(x, each(y, person(y), likes(x, y))) each(y, person(y), who(x, likes(x, y)))",
"uris": null
},
"FIGREF6": {
"num": null,
"type_str": "figure",
"text": "The derived syntax and semantics for Sentence 6.",
"uris": null
},
"FIGREF7": {
"num": null,
"type_str": "figure",
"text": "Alternative tree pairs for Mary and a that model topicalization. (8) A picture of Mary, John likes.a(x, and(picture(x), of(mary,x)), likes(x,john))",
"uris": null
}
}
}
} |