File size: 59,854 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 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 | {
"paper_id": "2020",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T13:27:27.167997Z"
},
"title": "TED-MDB Lexicons: Tr-EnConnLex, Pt-EnConnLex",
"authors": [
{
"first": "Murathan",
"middle": [],
"last": "Kurfal\u0131",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Stockholm University",
"location": {
"settlement": "Stockholm",
"country": "Sweden"
}
},
"email": "murathan.kurfali@ling.su.se"
},
{
"first": "Deniz",
"middle": [],
"last": "Zeyrek",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Middle East Technical University",
"location": {
"settlement": "Ankara",
"country": "Turkey"
}
},
"email": "dezeyrek@metu.edu.tr"
},
{
"first": "Am\u00e1lia",
"middle": [],
"last": "Mendes",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Middle East Technical University",
"location": {
"settlement": "Ankara",
"country": "Turkey"
}
},
"email": "amaliamendes@letras.ulisboa.pt"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "In this work, we present two new bilingual discourse connective lexicons, namely, for Turkish-English and European Portuguese-English created automatically using the existing discourse relation-aligned TED-MDB corpus. In their current form, the Pt-En lexicon includes 95 entries, whereas the Tr-En lexicon contains 133 entries. The lexicons constitute the first step of a larger project of developing a multilingual discourse connective lexicon.",
"pdf_parse": {
"paper_id": "2020",
"_pdf_hash": "",
"abstract": [
{
"text": "In this work, we present two new bilingual discourse connective lexicons, namely, for Turkish-English and European Portuguese-English created automatically using the existing discourse relation-aligned TED-MDB corpus. In their current form, the Pt-En lexicon includes 95 entries, whereas the Tr-En lexicon contains 133 entries. The lexicons constitute the first step of a larger project of developing a multilingual discourse connective lexicon.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "During the past decade or so the interest in discourse studies have dramatically increased following the release of the PDTB 2.0 corpus (Prasad et al., 2008) and, later, with the TextLink initiative 1 . In parallel to this interest, available resources annotated for various discourse-level phenomena have expanded, where discourse relational devices (DRDs) have received a special interest leading to the Connective-Lex database (Stede et al., 2019) . ConnLex is a joint online database project, which is the first attempt to bring together connective lexicons of different languages. It currently hosts the connective lexicons of nine different languages providing a web-based interface together with a crosslinguistically applicable XML schema. The entries in the lexicons provide fundamental information about discourse connectives, such as orthography, syntactic category, and their senses. The ConnLex project pursues the aim of expanding the database both in coverage (by adding new languages) and depth of the information. However, except for a few resources, most of the previous effort on devising discourse connective lexicons has relied on monolingual resources and any multilingual links that were provided have not gone beyond offering English equivalents. Few exceptions involve the bilingual Italian-German contrastive/concessive connective lexicon based on the cross-lingual projection of monolingual lexicons for Italian and German (Bourgonje et al., 2017) , and the very recent GeCzLex, Anaphoric Connective Lexicon for Czech and German (Pol\u00e1kov\u00e1 et al., 2020) .",
"cite_spans": [
{
"start": 136,
"end": 157,
"text": "(Prasad et al., 2008)",
"ref_id": "BIBREF8"
},
{
"start": 199,
"end": 200,
"text": "1",
"ref_id": null
},
{
"start": 430,
"end": 450,
"text": "(Stede et al., 2019)",
"ref_id": "BIBREF11"
},
{
"start": 1450,
"end": 1474,
"text": "(Bourgonje et al., 2017)",
"ref_id": "BIBREF0"
},
{
"start": 1556,
"end": 1579,
"text": "(Pol\u00e1kov\u00e1 et al., 2020)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The main contributions of the present study are (1) proposing an alternative way of producing bilingual lexicons, potentially applicable to building multilingual lexicons, (2) providing new bilingual discourse connective lexicons for European Portuguese-English and Turkish-English by (3) considering not only the explicit discourse relations but also the implicit relations in a recent multilingual discourse bank, namely TED-Multilingual Discourse Bank (TED-MDB) annotated in the PDTB style . The lexicon entries are extracted from TED-MDB, where each relation in the source language is aligned to its semantic equivalent in the target languages (Turkish and European Portuguese) (see \u00a72.1). In their current form, the Pt-En lexicon includes 95 entries covering 51 connectives in Portuguese and 57 connectives in English, while the Tr-En lexicon contains 133 entries with 72 connectives in Turkish and 56 in English.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The rest of the study is structured as follows: We firstly summarize the main data source, TED-MDB followed by the discourse relation alignment procedure ( \u00a72), the output of which is used as inputs to construct bilingual lexicons. \u00a73 describes the construction of the bilingual lexicons in detail. In \u00a74, we discuss issues concerning our lexicon construction procedure. \u00a75 concludes the paper presenting some future directions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "TED-MDB is a resource of TED talk transcripts comprising 7 languages manually annotated for discourse relations. It includes English, the source language (SL) along with transcribed texts in Ger-man, Lithuanian, European Portuguese, Russian, Turkish and Polish (target languages, or TLs). 2 Following the rules and principles of the PDTB, it annotates five discourse relations types (henceforth, DRs) with respect to the PDTB-3 sense hierarchy and ultimately aims to provide a clearly described level of discourse structure and semantics in multiple languages, thus engendering discourse parsing studies in multiple languages. TED-MDB currently involves 6 TED talk transcripts annotated with 5 DR types (Explicit, Implicit, AltLex, EntRel, NoRel), their senses and binary arguments, amounting to a total of 3649 tokens. The annotations have been carried out by native speaker annotators of the languages involved using the PDTB annotation tool. (Lee et al., 2016) 3 This tool stores the DR annotations in separate pipe-delimited files.",
"cite_spans": [
{
"start": 289,
"end": 290,
"text": "2",
"ref_id": null
},
{
"start": 945,
"end": 965,
"text": "(Lee et al., 2016) 3",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "TED-MDB",
"sec_num": "2"
},
{
"text": "To create TED-MDB, each monolingual team annotated the texts independently of the original texts to avoid the risk of the original language influencing the annotations. Yet, due to cross-lingual variation in rendering DRs, this design criterion led to tokens not existing in the original language . As the extraction of bilingual DC lexicons requires aligned relations, in the present study, our pipeline starts with the alignment of DRs following\u00d6zer and . Firstly, the DR annotations originally kept in pipe-delimited files were transferred onto the base text files of both TLs generating an ID for each. Then, wordand punctuation-tokenization as well as sentence alignment procedures were performed, followed by manual corrections of the latter. For DR alignment, all DRs in each bi-text unit were paired constructing DR matrices. The text pieces constituting discourse relations were translated into the SL using the Google Translate API and stop words were removed. Next, semantic similarity, taken in terms of cosine distance, was calculated between the source and target text segments using Word2Vec (Mikolov et al., 2013) within the range of 0 (\"no similarity\") to 1 (\"perfect similarity\"). DR pairs with a similarity over 0.7 were further evaluated for alignment.",
"cite_spans": [
{
"start": 1107,
"end": 1129,
"text": "(Mikolov et al., 2013)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Alignment Procedure",
"sec_num": "2.1"
},
{
"text": "For DR pairs with acceptable scores, the similarity of the DR sense and type was evaluated using a ranking algorithm which depends on the sense 2 https://github.com/MurathanKurfali/Ted-MDB-Annotations 3 https://www.cis.upenn.edu/ pdtb/annotator.html tags on the DRs. A score that reflected the SL-TL match was added to the semantic similarity score, where the DR type and the SL sense were both considered. The DR pair with the maximum score was marked as an aligned pair, and the same procedure was repeatedly applied until no DR pair was left in the matrices.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Alignment Procedure",
"sec_num": "2.1"
},
{
"text": "All the aligned pairs were manually checked by the authors.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Alignment Procedure",
"sec_num": "2.1"
},
{
"text": "The alignment algorithm has an F-score of 0.78 for Turkish-English and 0.81 for European Portuguese-English distributed over six documents accepting English annotations as the gold standard.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Alignment Procedure",
"sec_num": "2.1"
},
{
"text": "As shown in Pol\u00e1kov\u00e1 et al. (2020) and Bourgonje et al. (2017) , preparing a bilingual lexicon of discourse connectives is not a straightforward task requiring a variety of resources to compute a translation candidate table including monolingual DC lexicons of the TLs and a large parallel corpus (with at least 2M parallel sentences). A monolingual discourse connective lexicon exists for Portuguese (Mendes and Lejeune, 2016) and one is being developed for Turkish (Zeyrek and Ba\u015f\u0131b\u00fcy\u00fck, 2019) but parallel corpora of the required size are absent for the language pairs under investigation. Thus, the current study is built on the observation that just as monolingual lexicons can be compiled from annotated resources, bilingual dictionaries of discourse connectives can be constructed from a similar though low scaled parallel corpus such as TED-MDB. This corpus includes 375 bi-sentence units for English-Turkish and 364 for English-European Portuguese. The rest of the section describes the method employed to create two such bilingual DC lexicons of English-Turkish and English-European Portuguese.",
"cite_spans": [
{
"start": 12,
"end": 34,
"text": "Pol\u00e1kov\u00e1 et al. (2020)",
"ref_id": "BIBREF7"
},
{
"start": 39,
"end": 62,
"text": "Bourgonje et al. (2017)",
"ref_id": "BIBREF0"
},
{
"start": 401,
"end": 427,
"text": "(Mendes and Lejeune, 2016)",
"ref_id": "BIBREF3"
},
{
"start": 467,
"end": 495,
"text": "(Zeyrek and Ba\u015f\u0131b\u00fcy\u00fck, 2019)",
"ref_id": "BIBREF14"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "TED-MDB Lexicons",
"sec_num": "3"
},
{
"text": "Given the availability of TED-MDB, we propose an alternative way of building bilingual DC lexicons, which can be seen as the multilingual extension of extracting DC lexicons from annotated resources as in Mendes and del R\u00edo (2018); Das et al. (2018) .",
"cite_spans": [
{
"start": 232,
"end": 249,
"text": "Das et al. (2018)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "The method accepts a set of aligned DRs as input. For pre-processing, we firstly filter out all aligned pairs which contain a non-Explicit or non-Implicit relation followed by the removal of the pairs which are not annotated with exactly the same sense. This step helps us to eliminate the translation-based noise in the corpus as it is not uncommon for the senses of DRs to be lost or modified during translation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "After the pre-processing step, the bilingual lexicons are constructed in the following way:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "\u2022 For each connective in the SL, the list of senses in the input is computed.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "\u2022 The translation equivalents of the given connective are found in the TL using the aligned DRs. The translations are grouped under the senses found in the first step. Hence, we create different entries for each sense conveyed by the connective in SL. For example, in Tr-En, the \"but/ama\" pair appears both under the Comparison:Concession:Arg2-as-denier sense and the Comparison:Contrast sense.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "Due to the limited number of explicit DRs in TED-MDB (Table 1) , we also include in our lexicon implicit connectives which are the connectives inserted to implicit DRs by the annotators (Prasad et al., 2008 ). An inserted implicit connective can be regarded as the most suitable overt marker for a given implicit relation; hence, the pair of implicit connectives extracted from an aligned implicit DR is as valid an entry for our lexicon just as a pair of explicit connectives extracted from an aligned explicit DR. However, in order to keep things separated and facilitate further research, we create different entries for explicit and implicit usages of connectives in our lexicon. The detailed statistics about the lexicons are provided in Table 2 .",
"cite_spans": [
{
"start": 186,
"end": 206,
"text": "(Prasad et al., 2008",
"ref_id": "BIBREF8"
}
],
"ref_spans": [
{
"start": 53,
"end": 62,
"text": "(Table 1)",
"ref_id": "TABREF1"
},
{
"start": 743,
"end": 750,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Populating lexicon entries automatically",
"sec_num": "3.1"
},
{
"text": "The inspection of the automatically extracted connective pairs reveals several issues, which can mostly be attributed to translation strategies. In certain cases, translators use a completely different linguistic construction in the TL; yet, they manage to preserve the sense of the SL text (Example 1).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "Since both relations are annotated with the same sense, our method erroneously assumes these different connectives form a valid pair.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "(1) by investing sustainably, we're doing two things .. Quando investimos na sustentabilidade estamos a fazer duas coisas 'When we invest in sustainability, we are doing two things..'",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "In order to fix such cases, we firstly adopted a fully automatic approach where we tried to eliminate the unacceptable pairs by checking them against comprehensive bilingual dictionaries similar to Pol\u00e1kov\u00e1 et al. (2020) . To this end, we used Treq (\u0160krabal and Vav\u0159\u00edn, 2017) and the OPUS word alignment database. 4 However, both resources turned out to be unsuitable for our purposes. The translation candidate tables created from these resources eliminate a nontrivial amount of acceptable pairs as most of the time, valid translations are either absent in the databases or are assigned a very low probability, making it virtually impossible to determine an appropriate threshold between unacceptable and acceptable translations. That some of the Turkish connectives are suffixal connectives further render the use of dictionaries impractical. Therefore, we manually went through each entry in the lexicons in order to reach gold pairs. As the lexicons are not large and the task of deciding whether two words are translation equivalents is not too challenging, the manual control was completed within hours. The decision was made unanimously, which resulted in the removal of 9 pairs from Portuguese and only 2 from Turkish. It is also worth noting that the eliminated pairs overwhelmingly had the label Expansion:Level-of-detail:Arg2-asdetail, which \"is used when Arg2 describes in more detail, the situation in Arg1\" (Webber et al., 2019 Table 2 : Statistics regarding the constructed lexicons. \"Exp\" and \"Imp\" refers to the number of Explicit and Implicit connectives, respectively. The \"Total\" column represents the number of connectives when implicit and explicit connectives are counted as separate entries and when their type is disregarded (within parenthesis). The \"Monolingual\" column represents the number of connectives in the the respective language's monolingual lexicon (retrieved from (Stede et al., 2019) ) . The last column presents the minimum, maximum and the average number of translation equivalents in the target language.",
"cite_spans": [
{
"start": 198,
"end": 220,
"text": "Pol\u00e1kov\u00e1 et al. (2020)",
"ref_id": "BIBREF7"
},
{
"start": 1422,
"end": 1442,
"text": "(Webber et al., 2019",
"ref_id": "BIBREF13"
},
{
"start": 1904,
"end": 1924,
"text": "(Stede et al., 2019)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [
{
"start": 1443,
"end": 1450,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "A close examination showed that this subsense was not conveyed by the annotated DC tokens in the SL but rather inferred from the arguments, leading the translator to render the DR almost freely with a mismatching token in the TL. The removed pairs are as follows:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "\u2022 Pt-En: e -rather, e -for that matter, enquanto and, assim -that is, de facto -specifically, e as well as, e -lastly, isto\u00e9 -clearly, assimspecifically",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "\u2022 Tr-En:\u00f6zetleclearly, yani -clearly, i\u015fteclearly",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Post-process",
"sec_num": "3.2"
},
{
"text": "Each entry in the TED-MDB lexicons corresponds to a specific connective in the TL and a list of its possible translations in the TL grouped under the sense the connective conveys. Specifically, an entry consists of the following components (illustrated in Figure 1 ):",
"cite_spans": [],
"ref_spans": [
{
"start": 256,
"end": 264,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "\u2022 Connective: The head of each entry is a DC represented in its lemmatized form.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "\u2022 Dimlex link: Each DC and its translations are accompanied with an URL to their respective connective-lex entry, 5 which serves as a bridge between the bilingual and monolingual lexicons.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "\u2022 Sense list: The list of the senses that the head connective conveys in TED-MDB is displayed in the main screen of the interface sorted by the corpus frequencies of the senses.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "\u2022 List of translation candidates: For each sense in the list, the translation candidates 5 http://connective-lex.info/ specified in TL texts are provided. The translation candidates also have their own entries and are accessible just by clicking.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "\u2022 Example sentence: Each connective pair is accompanied with a randomly selected sentence pair from TED-MDB.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Structure of the Lexicons",
"sec_num": "3.3"
},
{
"text": "To the best of our knowledge, the TED-MDB lexicons presented here constitute the first attempt to construct a bilingual connective lexicon directly from an annotated parallel resource. Compilation of bilingual lexicons in this way has a number of practical benefits, where the main advantage is being not dependent on external resources. It alleviates the need for parallel corpora required to extract the translation candidates to map the connectives in different languages onto each other and does not necessitate monolingual DC lexicons, a challenging and time consuming effort especially when started from scratch (Roze et al., 2012) . Also, as all entries are populated from an annotated corpus, the lexicons are guaranteed to be symmetrical, and the bilingual examples provide an opportunity to observe the usage of connectives in context in two languages. It must also be noted that despite being compiled from a set of merely 300+ relations in each language set, our bilingual lexicons roughly account for 30% of the documented connectives of these languages; hence, their coverage is more impressive than it looks (Table 2) . As explained in Section 3.2, there are certain cases where the connectives from an aligned DR pair do not form valid lexicon entries. This issue revealed the larger problem that translational candidate tables, even those from a large parallel corpus like InterCorp (\u0160krabal and Vav\u0159\u00edn, 2017) cannot Figure 1 : The entry for 'also'in Tr-En lexicon adequately capture the translation equivalents of connectives. We believe this finding further highlights the need for such bilingual lexicons.",
"cite_spans": [
{
"start": 618,
"end": 637,
"text": "(Roze et al., 2012)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [
{
"start": 1123,
"end": 1132,
"text": "(Table 2)",
"ref_id": null
},
{
"start": 1434,
"end": 1442,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "Finally, as the manual control of the DR alignments constitutes one of the two non-automatic steps of our pipeline, we investigated its effect on the final lexicons to guide future research. To our surprise, the automatic alignment procedure turns out to be more than satisfactory: we were able to fetch more than 96% of all entries in the gold lexicons, suggesting that even a multilingual lexicon involving all languages in TED-MDB can be automatically constructed. This is left for future work.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "In translation, the choice of a DC that best conveys the sense of a relation and renders the relation in a natural way is not a trivial task. At a minimum, it requires careful consideration of the multiple senses of the connectives and their parts-of-speech. Even a bilingual dictionary is not always helpful for a translator. Bilingual lexicons built on the basis of naturalistic data is important to aid both machine and human translation as well as second language learners. In this study, we described a method of building two bilingual lexicons using aligned DR annotations. Both lexicons are available online as HTML web pages. 6 In contrast to previous bilingual lexicon studies, we did not use monolingual connective lexicons or dictionaries as the former was absent (at least for Turkish), and the latter caused loss of useful data. Although the alignment and the lexicon extraction procedures have been applied to two languages so far, this approach has the potential to be extended to other language pairs covered in the TED-MDB corpus, and this is what we plan to do as a future study. 6 http://metu-db.info/mdb/ted/resources.jsf",
"cite_spans": [
{
"start": 634,
"end": 635,
"text": "6",
"ref_id": null
},
{
"start": 1098,
"end": 1099,
"text": "6",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5"
},
{
"text": "http://opus.nlpl.eu/lex.php",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Toward a bilingual lexical database on connectives: Exploiting a german/italian parallel corpus",
"authors": [
{
"first": "Peter",
"middle": [],
"last": "Bourgonje",
"suffix": ""
},
{
"first": "Yulia",
"middle": [],
"last": "Grishina",
"suffix": ""
},
{
"first": "Manfred",
"middle": [],
"last": "Stede",
"suffix": ""
}
],
"year": 2017,
"venue": "Proceedings of the Fourth Italian Conference on Computational Linguistics-CLIC-IT",
"volume": "",
"issue": "",
"pages": "53--58",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Peter Bourgonje, Yulia Grishina, and Manfred Stede. 2017. Toward a bilingual lexical database on con- nectives: Exploiting a german/italian parallel corpus. In Proceedings of the Fourth Italian Conference on Computational Linguistics-CLIC-IT, pages 53-58.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Constructing a lexicon of english discourse connectives",
"authors": [
{
"first": "Debopam",
"middle": [],
"last": "Das",
"suffix": ""
},
{
"first": "Tatjana",
"middle": [],
"last": "Scheffler",
"suffix": ""
},
{
"first": "Peter",
"middle": [],
"last": "Bourgonje",
"suffix": ""
},
{
"first": "Manfred",
"middle": [],
"last": "Stede",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 19th Annual SIGDIAL Meeting on Discourse and Dialogue",
"volume": "",
"issue": "",
"pages": "360--365",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Debopam Das, Tatjana Scheffler, Peter Bourgonje, and Manfred Stede. 2018. Constructing a lexicon of en- glish discourse connectives. In Proceedings of the 19th Annual SIGDIAL Meeting on Discourse and Di- alogue, pages 360-365.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Annotating discourse relations with the pdtb annotator",
"authors": [
{
"first": "Alan",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Rashmi",
"middle": [],
"last": "Prasad",
"suffix": ""
},
{
"first": "Bonnie",
"middle": [],
"last": "Webber",
"suffix": ""
},
{
"first": "Aravind",
"middle": [],
"last": "Joshi",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: System Demonstrations",
"volume": "",
"issue": "",
"pages": "121--125",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alan Lee, Rashmi Prasad, Bonnie Webber, and Ar- avind Joshi. 2016. Annotating discourse relations with the pdtb annotator. In Proceedings of COLING 2016, the 26th International Conference on Compu- tational Linguistics: System Demonstrations, pages 121-125.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Ldm-pt-a portuguese lexicon of discourse markers",
"authors": [
{
"first": "Am\u00e1lia",
"middle": [],
"last": "Mendes",
"suffix": ""
},
{
"first": "Pierre",
"middle": [],
"last": "Lejeune",
"suffix": ""
}
],
"year": 2016,
"venue": "Conference Handbook of TextLink-Structuring Discourse in Multilingual Europe Second Action Conference",
"volume": "",
"issue": "",
"pages": "89--92",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Am\u00e1lia Mendes and Pierre Lejeune. 2016. Ldm-pt-a portuguese lexicon of discourse markers. In Confer- ence Handbook of TextLink-Structuring Discourse in Multilingual Europe Second Action Conference, pages 89-92. Debrecen University Press.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Using a discourse bank and a lexicon for the automatic identification of discourse connectives",
"authors": [
{
"first": "Am\u00e1lia",
"middle": [],
"last": "Mendes",
"suffix": ""
},
{
"first": "Iria",
"middle": [],
"last": "Del R\u00edo",
"suffix": ""
}
],
"year": 2018,
"venue": "International Conference on Computational Processing of the Portuguese Language",
"volume": "",
"issue": "",
"pages": "211--221",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Am\u00e1lia Mendes and Iria del R\u00edo. 2018. Using a dis- course bank and a lexicon for the automatic identi- fication of discourse connectives. In International Conference on Computational Processing of the Por- tuguese Language, pages 211-221. Springer.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Distributed representations of words and phrases and their compositionality",
"authors": [
{
"first": "Tomas",
"middle": [],
"last": "Mikolov",
"suffix": ""
},
{
"first": "Ilya",
"middle": [],
"last": "Sutskever",
"suffix": ""
},
{
"first": "Kai",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "Greg",
"middle": [
"S"
],
"last": "Corrado",
"suffix": ""
},
{
"first": "Jeff",
"middle": [],
"last": "Dean",
"suffix": ""
}
],
"year": 2013,
"venue": "Advances in neural information processing systems",
"volume": "",
"issue": "",
"pages": "3111--3119",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Cor- rado, and Jeff Dean. 2013. Distributed representa- tions of words and phrases and their compositional- ity. In Advances in neural information processing systems, pages 3111-3119.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "An automatic discourse relation alignment experiment on TED-MDB",
"authors": [
{
"first": "Deniz",
"middle": [],
"last": "Sibel\u00f6zer",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Zeyrek",
"suffix": ""
}
],
"year": 2019,
"venue": "Proceedings of the 2019 Workshop on Widening NLP",
"volume": "",
"issue": "",
"pages": "31--34",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sibel\u00d6zer and Deniz Zeyrek. 2019. An automatic dis- course relation alignment experiment on TED-MDB. In Proceedings of the 2019 Workshop on Widening NLP, pages 31-34, Florence, Italy. Association for Computational Linguistics.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Geczlex: Lexicon of czech and german anaphoric connectives",
"authors": [
{
"first": "Lucie",
"middle": [],
"last": "Pol\u00e1kov\u00e1",
"suffix": ""
},
{
"first": "Kate\u0159ina",
"middle": [],
"last": "Rysov\u00e1",
"suffix": ""
},
{
"first": "Magdal\u00e9na",
"middle": [],
"last": "Rysov\u00e1",
"suffix": ""
},
{
"first": "Ji\u0159\u00ed",
"middle": [],
"last": "M\u00edrovsk\u1ef3",
"suffix": ""
}
],
"year": 2020,
"venue": "Proceedings of The 12th Language Resources and Evaluation Conference",
"volume": "",
"issue": "",
"pages": "1089--1096",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lucie Pol\u00e1kov\u00e1, Kate\u0159ina Rysov\u00e1, Magdal\u00e9na Rysov\u00e1, and Ji\u0159\u00ed M\u00edrovsk\u1ef3. 2020. Geczlex: Lexicon of czech and german anaphoric connectives. In Proceedings of The 12th Language Resources and Evaluation Conference, pages 1089-1096.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "The penn discourse treebank 2.0",
"authors": [
{
"first": "Rashmi",
"middle": [],
"last": "Prasad",
"suffix": ""
},
{
"first": "Nikhil",
"middle": [],
"last": "Dinesh",
"suffix": ""
},
{
"first": "Alan",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Eleni",
"middle": [],
"last": "Miltsakaki",
"suffix": ""
},
{
"first": "Livio",
"middle": [],
"last": "Robaldo",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Aravind",
"suffix": ""
},
{
"first": "Bonnie",
"middle": [
"L"
],
"last": "Joshi",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Webber",
"suffix": ""
}
],
"year": 2008,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Milt- sakaki, Livio Robaldo, Aravind K Joshi, and Bon- nie L Webber. 2008. The penn discourse treebank 2.0. In LREC. Citeseer.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Lexconn: a french lexicon of discourse connectives. Discours. Revue de linguistique, psycholinguistique et informatique. A journal of linguistics, psycholinguistics and computational linguistics",
"authors": [
{
"first": "Charlotte",
"middle": [],
"last": "Roze",
"suffix": ""
},
{
"first": "Laurence",
"middle": [],
"last": "Danlos",
"suffix": ""
},
{
"first": "Philippe",
"middle": [],
"last": "Muller",
"suffix": ""
}
],
"year": 2012,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Charlotte Roze, Laurence Danlos, and Philippe Muller. 2012. Lexconn: a french lexicon of discourse connectives. Discours. Revue de linguistique, psy- cholinguistique et informatique. A journal of linguis- tics, psycholinguistics and computational linguistics, (10).",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "The translation equivalents database (treq) as a lexicographer's aid. In Electronic lexicography in the 21st century",
"authors": [
{
"first": "Martin",
"middle": [],
"last": "Michal\u0161krabal",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Vav\u0159\u00edn",
"suffix": ""
}
],
"year": 2017,
"venue": "Proceedings of eLex 2017 conference. Leiden: Lexical Computing",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Michal\u0160krabal and Martin Vav\u0159\u00edn. 2017. The transla- tion equivalents database (treq) as a lexicographer's aid. In Electronic lexicography in the 21st century. Proceedings of eLex 2017 conference. Leiden: Lexi- cal Computing.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Connective-lex: A web-based multilingual lexical resource for connectives. Discours. Revue de linguistique, psycholinguistique et informatique. A journal of linguistics, psycholinguistics and computational linguistics",
"authors": [
{
"first": "Manfred",
"middle": [],
"last": "Stede",
"suffix": ""
},
{
"first": "Tatjana",
"middle": [],
"last": "Scheffler",
"suffix": ""
},
{
"first": "Am\u00e1lia",
"middle": [],
"last": "Mendes",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Manfred Stede, Tatjana Scheffler, and Am\u00e1lia Mendes. 2019. Connective-lex: A web-based multilingual lexical resource for connectives. Discours. Revue de linguistique, psycholinguistique et informatique. A journal of linguistics, psycholinguistics and com- putational linguistics, (24).",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "A discourse-annotated corpus of conjoined vps",
"authors": [
{
"first": "Bonnie",
"middle": [],
"last": "Webber",
"suffix": ""
},
{
"first": "Rashmi",
"middle": [],
"last": "Prasad",
"suffix": ""
},
{
"first": "Alan",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Aravind",
"middle": [],
"last": "Joshi",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 10th Linguistic Annotation Workshop held in conjunction with ACL 2016",
"volume": "",
"issue": "",
"pages": "22--31",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bonnie Webber, Rashmi Prasad, Alan Lee, and Ar- avind Joshi. 2016. A discourse-annotated corpus of conjoined vps. In Proceedings of the 10th Linguistic Annotation Workshop held in conjunction with ACL 2016 (LAW-X 2016), pages 22-31.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "The penn discourse treebank 3.0 annotation manual",
"authors": [
{
"first": "Bonnie",
"middle": [],
"last": "Webber",
"suffix": ""
},
{
"first": "Rashmi",
"middle": [],
"last": "Prasad",
"suffix": ""
},
{
"first": "Alan",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Aravind",
"middle": [],
"last": "Joshi",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bonnie Webber, Rashmi Prasad, Alan Lee, and Ar- avind Joshi. 2019. The penn discourse treebank 3.0 annotation manual.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Tcl-a lexicon of turkish discourse connectives",
"authors": [
{
"first": "Deniz",
"middle": [],
"last": "Zeyrek",
"suffix": ""
},
{
"first": "Kezban",
"middle": [],
"last": "Ba\u015f\u0131b\u00fcy\u00fck",
"suffix": ""
}
],
"year": 2019,
"venue": "Proceedings of the First International Workshop on Designing Meaning Representations",
"volume": "",
"issue": "",
"pages": "73--81",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Deniz Zeyrek and Kezban Ba\u015f\u0131b\u00fcy\u00fck. 2019. Tcl-a lex- icon of turkish discourse connectives. In Proceed- ings of the First International Workshop on Design- ing Meaning Representations, pages 73-81.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Ted multilingual discourse bank (tedmdb): a parallel corpus annotated in the pdtb style",
"authors": [
{
"first": "Deniz",
"middle": [],
"last": "Zeyrek",
"suffix": ""
},
{
"first": "Am\u00e1lia",
"middle": [],
"last": "Mendes",
"suffix": ""
},
{
"first": "Yulia",
"middle": [],
"last": "Grishina",
"suffix": ""
},
{
"first": "Murathan",
"middle": [],
"last": "Kurfal\u0131",
"suffix": ""
},
{
"first": "Samuel",
"middle": [],
"last": "Gibbon",
"suffix": ""
},
{
"first": "Maciej",
"middle": [],
"last": "Ogrodniczuk",
"suffix": ""
}
],
"year": 2019,
"venue": "Language Resources and Evaluation",
"volume": "",
"issue": "",
"pages": "1--27",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Deniz Zeyrek, Am\u00e1lia Mendes, Yulia Grishina, Mu- rathan Kurfal\u0131, Samuel Gibbon, and Maciej Ogrod- niczuk. 2019. Ted multilingual discourse bank (ted- mdb): a parallel corpus annotated in the pdtb style. Language Resources and Evaluation, pages 1-27.",
"links": null
}
},
"ref_entries": {
"TABREF1": {
"content": "<table/>",
"text": "Distribution of discourse relation types in TED-MDB",
"html": null,
"type_str": "table",
"num": null
},
"TABREF2": {
"content": "<table><tr><td>Language</td><td colspan=\"4\"># of Connectives Exp Imp Total (Unique) Monolingual</td><td># of Sense</td><td colspan=\"3\"># of Translations Min Max Avg</td></tr><tr><td>English</td><td>26</td><td>31</td><td>57 (48)</td><td>142</td><td>1.23</td><td>1</td><td>6</td><td>1.36</td></tr><tr><td colspan=\"2\">Portuguese 26</td><td>25</td><td>51 (42)</td><td>-</td><td>1.49</td><td>1</td><td>4</td><td>1.36</td></tr><tr><td>English</td><td>24</td><td>32</td><td>56 (47)</td><td>142</td><td>1.29</td><td>1</td><td>7</td><td>1.83</td></tr><tr><td>Turkish</td><td>34</td><td>38</td><td>72 (62)</td><td>226</td><td>1.44</td><td>1</td><td>4</td><td>1.26</td></tr><tr><td/><td/><td/><td>151</td><td/><td/><td/><td/><td/></tr></table>",
"text": ").",
"html": null,
"type_str": "table",
"num": null
}
}
}
} |