File size: 108,731 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 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 | {
"paper_id": "P13-1027",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T09:32:25.273316Z"
},
"title": "Translating Italian connectives into Italian Sign Language",
"authors": [
{
"first": "Camillo",
"middle": [],
"last": "Lugaresi",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Illinois at Chicago Politecnico di Milano",
"location": {}
},
"email": ""
},
{
"first": "Barbara",
"middle": [],
"last": "Di",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "University of Illinois at Chicago",
"location": {}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "We present a corpus analysis of how Italian connectives are translated into LIS, the Italian Sign Language. Since corpus resources are scarce, we propose an alignment method between the syntactic trees of the Italian sentence and of its LIS translation. This method, and clustering applied to its outputs, highlight the different ways a connective can be rendered in LIS: with a corresponding sign, by affecting the location or shape of other signs, or being omitted altogether. We translate these findings into a computational model that will be integrated into the pipeline of an existing Italian-LIS rendering system. Initial experiments to learn the four possible translations with Decision Trees give promising results.",
"pdf_parse": {
"paper_id": "P13-1027",
"_pdf_hash": "",
"abstract": [
{
"text": "We present a corpus analysis of how Italian connectives are translated into LIS, the Italian Sign Language. Since corpus resources are scarce, we propose an alignment method between the syntactic trees of the Italian sentence and of its LIS translation. This method, and clustering applied to its outputs, highlight the different ways a connective can be rendered in LIS: with a corresponding sign, by affecting the location or shape of other signs, or being omitted altogether. We translate these findings into a computational model that will be integrated into the pipeline of an existing Italian-LIS rendering system. Initial experiments to learn the four possible translations with Decision Trees give promising results.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Automatic translation between a spoken language and a signed language gives rise to some of the same difficulties as translation between spoken languages, but adds unique challenges of its own. Contrary to what one might expect, sign languages are not artificial languages, but natural languages that spontaneously arose within deaf communities; although they are typically named after the region where they are used, they are not derived from the local spoken language and tend to bear no similarity to it. Therefore, translation from any spoken language into the signed language of that specific region is at least as complicated as between any pairs of unrelated languages.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The problem of automatic translation is compounded by the fact that the amount of computational resources to draw on is much smaller than is typical for major spoken languages. Moreover, the fact that sign languages employ a different transmission modality (gestures and expressions instead of sounds) means that existing writing systems are not easily adaptable to them. The resulting lack of a shared written form does nothing to improve the availability of sign language corpora; bilingual corpora, which are of particular importance to a translation system, are especially rare. In fact, various projects around the world are trying to ameliorate this sad state of affairs for specific Sign Languages (Lu and Huenerfauth, 2010; Braffort et al., 2010; Morrissey et al., 2010) .",
"cite_spans": [
{
"start": 705,
"end": 731,
"text": "(Lu and Huenerfauth, 2010;",
"ref_id": "BIBREF20"
},
{
"start": 732,
"end": 754,
"text": "Braffort et al., 2010;",
"ref_id": "BIBREF4"
},
{
"start": 755,
"end": 778,
"text": "Morrissey et al., 2010)",
"ref_id": "BIBREF26"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we describe the work we performed as concerns the translation of connectives from the Italian language into LIS, the Italian Sign Language (Lingua Italiana dei Segni). Because the communities of signers in Italy are relatively small and fragmented, and the language has a relatively short history, there is far less existing research and material to draw on than for, say, ASL (American Sign Language) or BSL (British Sign Language).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Our work was undertaken within the purview of the ATLAS project (Bertoldi et al., 2010; Lombardo et al., 2011; Prinetto et al., 2011; Mazzei, 2012; Ahmad et al., 2012) , which developed a full pipeline for translating Italian into LIS. ATLAS is part of a recent crop of projects devoted to developing automatic translation from language L spoken in geographic area G into the sign language spoken in G (Dreuw et al., 2010; L\u00f3pez-Lude\u00f1a et al., 2011; Almohimeed et al., 2011; Lu and Huenerfauth, 2012) . Input is taken in the form of written Italian text, parsed, and converted into a semantic representation of its contents; from this semantic representation, LIS output is produced, using a custom serialization format called AEWLIS (which we will describe later). This representation is then augmented with space positioning information, and fed into a final renderer component that performs the signs using a virtual actor. ATLAS focused on a limited domain for which a bilingual Italian/LIS cor-pus was available: weather forecasts, for which the Italian public broadcasting corporation (RAI) had long been producing special broadcasts with a signed translation. This yielded a corpus of 376 LIS sentences with corresponding Italian text: this corpus, converted into AEWLIS format, was the main data source for the project. Still, it is a very small corpus, hence the main project shied away from statistical NLP techniques, relying instead on rule-based approaches developed with the help of a native Italian/LIS bilingual speaker; a similar approach is taken e.g. in (Almohimeed et al., 2011) for Arabic.",
"cite_spans": [
{
"start": 64,
"end": 87,
"text": "(Bertoldi et al., 2010;",
"ref_id": "BIBREF3"
},
{
"start": 88,
"end": 110,
"text": "Lombardo et al., 2011;",
"ref_id": "BIBREF16"
},
{
"start": 111,
"end": 133,
"text": "Prinetto et al., 2011;",
"ref_id": "BIBREF27"
},
{
"start": 134,
"end": 147,
"text": "Mazzei, 2012;",
"ref_id": "BIBREF24"
},
{
"start": 148,
"end": 167,
"text": "Ahmad et al., 2012)",
"ref_id": "BIBREF0"
},
{
"start": 402,
"end": 422,
"text": "(Dreuw et al., 2010;",
"ref_id": "BIBREF8"
},
{
"start": 423,
"end": 449,
"text": "L\u00f3pez-Lude\u00f1a et al., 2011;",
"ref_id": "BIBREF17"
},
{
"start": 450,
"end": 474,
"text": "Almohimeed et al., 2011;",
"ref_id": "BIBREF1"
},
{
"start": 475,
"end": 500,
"text": "Lu and Huenerfauth, 2012)",
"ref_id": "BIBREF21"
},
{
"start": 1573,
"end": 1598,
"text": "(Almohimeed et al., 2011)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The main semantic-bearing elements of an Italian sentence, such as nouns or verbs, typically have a LIS sign as their direct translation. We focus on a different class of elements, comprising conjunctions and prepositions, but also some adverbs and prepositional phrases; collectively, we refer to them as connectives. Since they are mainly structural elements, they are more heavily affected by differences in the syntax and grammar of Italian and LIS (and, presumably, in those of any spoken language and the \"corresponding\" SL). Specifically, as we will see later, some connectives are translated with a sign, some connectives are dropped, whereas others affect the positioning of other signs, or just their syntactic proximity.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Why connectives?",
"sec_num": "1.1"
},
{
"text": "It should be noted that our usage of the term \"connectives\" is somewhat unorthodox. For example, while prepositions can be seen as connectives (Ferrari, 2008) , only a few adverbs can work as connectives. From the Italian Treebank, we extracted all words or phrases that belonged to a syntactic category that can be a connective (conjunction, preposition, adverb or prepositional phrase). We then found that we could better serve the needs of ATLAS by running our analysis on the entire resulting list, without filtering it by eliminating the entries that are not actual connectives. In fact, semantic differences re-emerge through our analysis: e.g., the temporal adverbs \"domani\" and \"dopodomani\" are nearly always preserved, as they do carry key information (especially for weather forecasting) and are not structural elements.",
"cite_spans": [
{
"start": 143,
"end": 158,
"text": "(Ferrari, 2008)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Why connectives?",
"sec_num": "1.1"
},
{
"text": "In performing our analysis, we pursued a different path from the main project, relying entirely on the bilingual corpus. Although the use of sta-tistical techniques was hampered by the small size of the corpus, at the same time it presented an interesting opportunity to attack the problem from a different angle. In this paper we describe how we uncovered the translation distributions of the different connectives from Italian to LIS via tree alignment.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Why connectives?",
"sec_num": "1.1"
},
{
"text": "The corpus consists of 40 weather forecasts in Italian and LIS. The Italian spoken utterance and LIS signing were transcribed from the original videos -one example of an Italian sentence and its LIS equivalent are shown in Figure 1 . An English word-by-word translation is provided for the Italian sentence, followed by a more fluent translation; the LIS glosses are literally translated. Note that as concerns LIS, this simply includes the gloss for the corresponding sign. The 40 weather forecast comprise 374 Italian sentences and 376 LIS sentences, stored in 372 AEWLIS files. In most cases, a file corresponds to one Italian sentence and one corresponding LIS sentences; however, there are 4 files where an Italian sentence is split into two LIS sentences, and 2 files where two Italian sentences are merged into one LIS sentence.",
"cite_spans": [],
"ref_spans": [
{
"start": 223,
"end": 231,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Corpus Analysis",
"sec_num": "2"
},
{
"text": "AEWLIS is an XML-based format (see Figure 2) which represents each sign in the LIS sentence as an element, in the order in which they occur in the sentence. A sign's lemma is represented by the Italian word with the same meaning, always written in uppercase, and with its part of speech (tipoAG in Figure 2 ); there are also IDs referencing the lemma's position in a few dictionaries, but these are not always present. The AEWLIS file also stores several additional attributes, such as: a parent reference that represents the syntax of the LIS sentence; the syntactic role \"played\" by the sign in the LIS sentence; the facial expression accompanying the gesture; the location in the signing space (which may be an absolute location or a reference to a previous sign's: compare HR (High Right) and atLemma in Figure 2 ). These attributes are stored as elements grouped by type, and reference the corresponding sign element by its ordinal position in the sentence. The additional attributes are not always available: morphological variations are annotated only when they differ from an assumed standard form of the sign, while the syntactic structure was annotated for only 89 sentences. Figure 1 : Italian sentence and its LIS translation <Lemmi> <NuovoLemma lemma=\"POMERIGGIO\" tipoAG=\"NOME\" ... endTime=\"2.247\" idSign=\"\"/> <NuovoLemma lemma=\"sardegna\" tipoAG=\"NOME_PROPRIO\" ... endTime=\"2.795\" idSign=\"2687\"/> <NuovoLemma lemma=\"area\" tipoAG=\"NOME\" ... endTime=\"4.08\" idSign=\"2642\"/> <NuovoLemma lemma=\"nuvola\" tipoAG=\"NOME\" ... endTime=\"5.486\" idSign=\"2667\"/> <NuovoLemma lemma=\"pure\" tipoAG=\"AVVERBIO\" ... endTime=\"6.504\" idSign=\"2681\"/> ... </Lemmi><SentenceAttribute> <Parent> <Timestamp time=\"1\" value=\"ID:3\"/> <Timestamp time=\"2\" value=\"ID:2\"/> <Timestamp time=\"3\" value=\"ID:3\"/> <Timestamp time=\"4\" value=\"root_1\"/> <Timestamp time=\"5\" value=\"ID:3\"/> ... </Parent> ... <Sign_Spatial_Location> <Timestamp time=\"1\" value=\"\"/> <Timestamp time=\"2\" value=\"HR\"/> <Timestamp time=\"3\" value=\"HL\"/> <Timestamp time=\"4\" value=\"atLemma(ID:2, Distant)\"/> <Timestamp time=\"5\" value=\"\"/> ... </Sign_Spatial_Location> ... <Facial> <Timestamp time=\"1\" value=\"\"/> <Timestamp time=\"2\" value=\"eye brows:raise\"/> <Timestamp time=\"3\" value=\"eye brows:raise\"/> <Timestamp time=\"4\" value=\"eye brows:-lwrd\"/> <Timestamp time=\"5\" value=\"\"/> ... </Facial> </SentenceAttribute> ",
"cite_spans": [],
"ref_spans": [
{
"start": 35,
"end": 41,
"text": "Figure",
"ref_id": null
},
{
"start": 298,
"end": 306,
"text": "Figure 2",
"ref_id": "FIGREF0"
},
{
"start": 808,
"end": 816,
"text": "Figure 2",
"ref_id": "FIGREF0"
},
{
"start": 1186,
"end": 1194,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Corpus Analysis",
"sec_num": "2"
},
{
"text": "The list of Italian connectives we considered was extracted from the Italian Treebank developed at the Institute for Computational Linguistics in Pisa, Italy (Montemagni et al., 2003) by searching for conjunctions, prepositions and adverbs. This yielded a total of 777 potential connectives. Of those, only 104 occur in our corpus. A simple count of the occurrences of connectives in the Italian and LIS versions of the corpus yields the following results:",
"cite_spans": [
{
"start": 158,
"end": 183,
"text": "(Montemagni et al., 2003)",
"ref_id": "BIBREF25"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(a) 78 connectives (2068 occurrences total) only occur in the Italian version, for example AL-MENO (at least), CON (with), INFATTI (indeed), PER (for) .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(b) 8 connectives (67 occurrences total) only occur in the LIS version, for example CIRCA (about), as in \"Here I am\"), PURE (also, additionally).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(c) 25 connectives (925 occurrences total) occur in both versions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "For the third category, we have computed the ratio of the number of occurrences in Italian over the number of occurrences in LIS; the ratios are plotted in logarithmic scale in Figure 3 . 0 on the scale corresponds to an ITA/LIS ratio equal to 1; positive numbers indicate that there are more occurrences in ITA, negative numbers that there are more occurrences in LIS. We can recognize three clusters by ratio:",
"cite_spans": [],
"ref_spans": [
{
"start": 177,
"end": 185,
"text": "Figure 3",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(c1) 9 connectives occurring in both languages, but mainly in Italian, for example POCO (a little), PI\u00da (more), SE (if), QUINDI (hence).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(c2) 13 connectives occurring in both languages with similar frequency, for example SOLO (only), POI (then), O (or), MA (but).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "(c3) 3 connectives occurring in both languages, but mainly in LIS: MENO (less), ADESSO (now), INVECE (instead). 3 The effect of the Italian connectives on the LIS translation",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "From this basic frequency analysis we can already notice that a large number of connectives only appear in Italian, or have far more occurrences in Italian than in LIS. This is unsurprising, considering that LIS sentences tend to be shorter than Italian sentences in terms of number of signs/words (a fact which probably correlates with the increased energy and time requirements intrinsic into articulating a message using one's arms rather than one's tongue). However, our goal is to predict when a connective should be dropped and when it should be preserved. Furthermore, even if the connective does not appear in the LIS sentence as a directly corresponding sign, that does not mean that its presence in the Italian sentence has no effect on the translation. We hypothesize four different possible realizations for a connective in the Italian sentence:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "\u2022 the connective word or phrase may map to a corresponding connective sign;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "\u2022 the connective is not present as a sign, but may affect the morphology of the signs which translate words syntactically adjacent to the connective;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "\u2022 the connective is not present as a sign, but its presence may be reflected by the fact that words connected by it map to signs which are close to each other in the LIS syntax tree;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "\u2022 the connective is dropped altogether.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "The second hypothesis deserves some explanation. The earliest treatments of LIS assumed that each sign (lemma) could be treated as invariant. Attempts to represent LIS in writing simply replaced each sign with a chosen Italian word (or phrase, if necessary) with the same meaning. Although this is still a useful way of representing the basic lemma, more recent studies have noted that LIS signs can undergo significant morphological variations which are lost under such a scheme. The AEWLIS format, in fact, was designed to preserve them.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "Of course, morphological variations in LIS are not phonetic, like in a spoken language, but gestural (Volterra, 1987; Romeo, 1991) . For example, the location in which a gesture is performed may be varied, or its speed, or the facial expressions that accompany it (Geraci et al., 2008) . One particularly interesting axis of morphology is the positioning of the gesture in the signing space in front of the signer. This space is implicitly divided into a grid with a few different positions from left to right and from top to bottom (see HR -High Right, and LH -High Left, in Figure 2 ). Two or more signs can then be placed in different positions in this virtual space, and by performing other signs in the same positions the signer can express a backreference to the previously established entity at that location. One can even have a movement verb where the starting and ending positions of the gesture are positioned independently to indicate the source and destination of the movement. In other words, these morphological variations can perform a similar function to gender and number agreement in Italian backreferences, but they can also assume roles that in Italian would be performed by prepositions, which are connectives. In fact, as we will see later on, Italian prepositions are never translated as signs, but are often associated with morphological variations on related signs.",
"cite_spans": [
{
"start": 101,
"end": 117,
"text": "(Volterra, 1987;",
"ref_id": "BIBREF29"
},
{
"start": 118,
"end": 130,
"text": "Romeo, 1991)",
"ref_id": "BIBREF28"
},
{
"start": 264,
"end": 285,
"text": "(Geraci et al., 2008)",
"ref_id": "BIBREF12"
}
],
"ref_spans": [
{
"start": 576,
"end": 584,
"text": "Figure 2",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Distributional statistics for connectives",
"sec_num": "2.1"
},
{
"text": "Two of our four translation hypotheses involve a notion of distance on the syntax tree, and a no- tion of signs corresponding to words. Therefore, it is not sufficient to consider the LIS sentence and the Italian sentence separately. Instead, their syntax trees must be reconstructed and aligned. Tree alignment in a variety of forms has been extensively used in machine translation systems (Gildea, 2003; Eisner, 2003; May and Knight, 2007) . As far as we know, we are the first to attempt the usage of tree alignment to aid in the translation between a spoken and a sign language, partly because corpora that include synctactic trees for sign language sentences hardly exist. (L\u00f3pez-Lude\u00f1a et al., 2011) does use alignment techniques for translation from Spanish to Spanish Sign Language (SSL), but it is limited to alignment between words or phrases in Spanish, and glosses or sequences of glosses in SSL.",
"cite_spans": [
{
"start": 391,
"end": 405,
"text": "(Gildea, 2003;",
"ref_id": "BIBREF13"
},
{
"start": 406,
"end": 419,
"text": "Eisner, 2003;",
"ref_id": "BIBREF9"
},
{
"start": 420,
"end": 441,
"text": "May and Knight, 2007)",
"ref_id": "BIBREF23"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "We have developed a pipeline that takes in input the corpus files, parses the Italian sentence with an existing parser, and retrieves / builds a parse tree for the LIS sentence. The two trees are then aligned by exploiting the word/sign alignment. A sample output is shown in Figure 4 .",
"cite_spans": [],
"ref_spans": [
{
"start": 276,
"end": 284,
"text": "Figure 4",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "Italian sentence parsing. Since the corpus contains the Italian sentences in plain, unstructured text form, they need to be parsed. We used the DeSR parser, a dependency parser pre-trained on a very large Italian corpus (Attardi et al., 2007; Ciaramita and Attardi, 2011) . This parser produced the syntax trees and POS tagging that we used for the Italian part of the corpus.",
"cite_spans": [
{
"start": 220,
"end": 242,
"text": "(Attardi et al., 2007;",
"ref_id": "BIBREF2"
},
{
"start": 243,
"end": 271,
"text": "Ciaramita and Attardi, 2011)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "LIS syntax tree. One of the attributes allowed by AEWLIS is \"parent\", which points a sign to its parent in the syntax tree, or marks it as a root (see Figure 2 ). These hand-built syntax trees are available in roughly 1/4 of the AEWLIS files. Because the size of our corpus is already limited, and be-cause no tools are available to generate LIS syntax trees, for the remaining 3/4 of the corpus we fell back on a simple linear tree where each sign is connected to its predecessor. This solution at least maintains locality in most cases.",
"cite_spans": [],
"ref_spans": [
{
"start": 151,
"end": 159,
"text": "Figure 2",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "Word Alignment. Having obtained syntax trees for the two sentences, we then needed to align them. For this purpose we used the Berkeley Word Aligner (BWA) 1 (Denero, 2007) , a general tool for aligning sentences in bilingual corpora. BWA takes as input a series of matching sentences in two different languages, trains multiple unsupervised alignment models, and selects the optimal result using a testing set of manual alignments. The output is a list of aligned word indices for each input sentence pair. On our data set, BWA performance is as follows: Precision = 0.736364, Recall = 0.704348, AER = 0.280000.",
"cite_spans": [
{
"start": 157,
"end": 171,
"text": "(Denero, 2007)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "Integration. The result is an integrated syntax tree representation of the Italian and LIS versions of the sentence, with arcs bridging aligned word/sign pairs. Since some connectives consist of multi-word phrases, the word nodes which are part of one are merged into a super-node that inherits all connections to other nodes. Figure 4 shows the end result for the Italian and LIS sentences in Figure 1 (the two sentences are repeated for convenience at the bottom of Figure 4 ). The rectangular boxes are words in the Italian sentence, while the rounded boxes are signs in the LIS sentence. The Italian tree has its root(s) at the bottom, while the LIS tree has its root(s) at the top. Solid arrows point from children to parent nodes in the syntax tree. Gray-shaded boxes represent connectives (words or signs, as indicated by the border of the box). Bold dashed lines show word alignment. Edges with round heads show relationships where a sign has a location attribute referencing another sign. Arrows with an empty triangular head trace the paths described in the next section.",
"cite_spans": [],
"ref_spans": [
{
"start": 327,
"end": 335,
"text": "Figure 4",
"ref_id": "FIGREF2"
},
{
"start": 394,
"end": 402,
"text": "Figure 1",
"ref_id": null
},
{
"start": 468,
"end": 476,
"text": "Figure 4",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "Tree Alignment",
"sec_num": "3.1"
},
{
"text": "At this point individual words are aligned, but that is not sufficient. Our hypotheses on the effect of connectives on translation requires us to align a tree fragment surrounding the Italian connective with the corresponding tree fragment on the LIS side -where the connective may be missing. In effect, since we have hypothesized that the presence of a connective can affect the translation of the two subtrees that it connects, we would like to be able to align each of those subtrees to its translation. However, given the differences between the two languages, it is not easy to give a clear definition of this mapping -let alone to compute it.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "Instead, we can take a step back to word-level alignment. We make the observation that, if two words belong to two different subtrees linked by a connective, so that the path between the two words goes through the connective, then the frontier between the LIS counterparts of those two subtrees should also lie along the path between the signs aligned with those two words. If the connective is preserved in translation as a sign, we should expect to find it along that path; if it is not, its effect should still be seen along that path, either in the form of morphological variations to the signs along the path, or in the shortness of the path itself.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The first step, then, is to split the Italian syntax tree by removing the connective. This yields one subtree containing the connective's parent node, if any, and one subtree for each of the connective's children, if any. The parent subtree typically contains most of the rest of the sentence, so only the direct ancestors of the connective are considered. Then, each pair of words belonging to different subtrees is linked by a path that goes through the connective in the original tree. Of these words, we select the ones that have aligned signs, and then we compute the path between each pair of signs aligned to words belonging to different subtrees. This gives us a set of paths to consider in the LIS syntax tree.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "For example, let us consider the connective \"di\" between \"possibilit\u00e0\" and \"scroscio\" in Figure 4 .",
"cite_spans": [],
"ref_spans": [
{
"start": 89,
"end": 97,
"text": "Figure 4",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 This node connects two subtrees: a child subtree containing \"qualche, breve, scroscio, di, pioggia\", and a parent subtree containing the rest of the sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 From each subtree, a set of paths is generated: all paths extending from the connective to the leaves of the child subtree (for example \"scroscio, qualche\" or \"scroscio, di, pioggia\"), and the path of direct ancestors in the parent tree (\"sulla, annuvolamento, possibilit\u00e0\").",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 Iterate through the cartesian product of each 22 10 (45.45%) 2 (9.09%) 0 (0.00%) 10 (45.45%) abbastanza 11 4 (36.36%) 1 (9.09%) 0 (0.00%) 6 (54.55%) anche 89 33 (37.08%) 5 (5.62%) 1 (1.12%) 53 (59.55%) ora 17 6 (35.29%) 1 (5.88%) 1 (5.88%) 10 (58.82%) proprio 11 5 (45.45%) 0 (0.00%) 0 (0.00%) 6 (54.55%) quindi 35 9 (25.71%) 1 (2.86%) 0 (0.00%) 25 (71.43%) come 16 0 (0.00%) 1 (6.25%) 1 (6.25%) 14 (87.50%) dove 28 0 (0.00%) 1 (3.57%) 0 (0.00%) 27 (96.43%) generalmente 13 0 (0.00%) 0 (0.00%) 0 (0.00%) 13 (100.00%) per quanto riguarda 14 0 (0.00%) 0 (0.00%) 1 (7.14%) 13 (92.86%) piuttosto 13 0 (0.00%) 0 (0.00%) 0 (0.00%) 13 (100.00%) pi\u00f9 57 0 (0.00%) 3 (5.26%) 2 (3.51%) 52 (91.23%) poco 63 2 (3.17%) 3 (4.76%) 0 (0.00%) 58 (92.06%) sempre 13 1 (7.69%) 0 (0.00%) 0 (0.00%) 12 (92.31%) soprattutto 16 1 (6.25%) 1 (6.25%) 0 (0.00%) 14 (87.50%) a 111 0 (0.00%) pair of sets (in this case we have only one pair), and consider the full path formed by the two paths connected by the connective node (for instance, \"sulla, annuvolamento, possiblit\u00e0, di, scroscio, breve\").",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 For each of these paths, take the signs aligned to words on different sides of the target connective, and find the shortest path between those signs in the LIS syntax tree; we call this the aligned path. For example, from \"possibilit\u00e0\" and \"scroscio\" we find \"POTERE, ACQUAZZONE\". If this process generates multiple paths, only the maximal ones are kept.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "By looking at words within a certain distance of the connective, at their aligned signs, and at the distance between those signs in the aligned path, the program then produces one or more \"translation candidates\" for each occurrence of a connective:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 Sign: if the connective word is aligned to a connective sign in LIS, that is its direct translation;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 Location: if morphology variations (currently limited to the \"location\" attribute, see Figure 2 ) are present on a sign aligned to an It. word belonging to one of the examined paths, and the word is less than 2 steps away from the connective, that morphological variation in LIS may capture the function of the connective;",
"cite_spans": [],
"ref_spans": [
{
"start": 89,
"end": 97,
"text": "Figure 2",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 Close: if two It. words are connected by a connective, and they map to signs which have a very short path between them (up to 3 nodes, including the two signs), the connective may be reflected simply in this close connection between the translated subtrees in the LIS syntax tree;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "\u2022 Missing: if none of the above hypotheses are possible, we hypothesize that the connective has been simply dropped. Table 1 shows the results of this analysis. It includes only connectives with more than 10 occurrences. For each connective and translation hypothesis, the shading of the cell is proportional to the fraction of occurrences where that hypothesis is possible; this fraction is also given as a percent. Note that Sign, Location and Close candidates are not mutually exclusive: for instance, an occurrence of a connective might be directly aligned with a sign, but at the same time it might fit the criteria for a Location candidate. For this reason, the sum of the percents in the four columns is not necessarily 100. k-means clustering (MacQueen, 1967; Lloyd, 1982) has been applied to the connectives, with the aforementioned fractions as the features. The resulting five clusters are represented by the row groupings in the table.",
"cite_spans": [
{
"start": 751,
"end": 767,
"text": "(MacQueen, 1967;",
"ref_id": "BIBREF22"
},
{
"start": 768,
"end": 780,
"text": "Lloyd, 1982)",
"ref_id": "BIBREF14"
}
],
"ref_spans": [
{
"start": 117,
"end": 124,
"text": "Table 1",
"ref_id": "TABREF2"
}
],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The first cluster contains words which clearly have a corresponding sign in LIS, such as \"domani\" (tomorrow). \"Domani\" and \"dopodomani\" are not actually connectives, while \"mentre\", \"o\" and \"per\u00f2\" are. It is interesting to note that, while a logician might expect \"e\" (and) and \"o\" (or) to be treated similarly, they actually work quite differently in LIS: there is a specific sign for \"o\", but there is no sign for \"e\". Instead, signs are simply juxtaposed in LIS where \"e\" would be used in Italian.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The words in the second cluster also have a direct sign translation, but they are missing in the LIS translation around half of the time. Several words represent connections with previous statements or situations, such as \"ancora\" (again), \"invece\" (instead), \"ma\" (but). These appear to be often dropped in LIS when they reference a previous sentence, e.g. a sentence-initial \"ma\"; or when they are redundant in Italian, e.g. \"ma\" in \"ma anche\" (\"but also\"). Therefore, we think can see two phenomena at play here: a stronger principle of economy in LIS, and a reduced number of explicit connections across sentences.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The third cluster is similar to the second cluster, but with a higher percent of dropped connectives. This is probably related to the semantics of these five words. \"Abbastanza\" means \"quite, enough\", and in general indicates a medium quantity, not particularly large nor particularly small. It is no surprise that this word is more likely to succumb to principles of economy in language. \"Anche\" means \"also\", and is either translated as \"PURE\" (also) or dropped. This does not seem to depend on the specific circumstances of its usage; rather, it seems to be largely a stylistic choice by the translator. \"Proprio\" (\"precisely\", \"just\") has a corre-sponding sign \"PROPRIO\", but since it does not convey essential information it is a good candidate for dropping. \"Quindi\", meaning \"therefore\", has its own sign \"QUINDI\", but once again the causal relationship it conveys is usually not essential to understanding what the weather will be, and thus it is frequently dropped.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The fourth cluster consists of connectives which are largely simply dropped. Some of these are elements that just contribute to the discourse flow in Italian, such as \"per quanto riguarda\" (\"concerning\"); in fact, this connective mainly occurs in sentence-initial position in the Italian sentences in our corpus and denotes a change of topic from the previous sentence, corroborating our hypothesis of a reduced use of explicit intersentence connections in LIS. It may seem strange for comparative and intensity markers such as \"pi\u00f9\" (more) or \"poco\" (a little) to be so consistently dropped, but it turns out that intensity variations for weather phenomena are often embedded into a specific sign, for example \"NUVOLOSIT\u00c0 AUMENTARE\" (increasing cloud cover).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "The fifth cluster contains all Italian prepositions (with 10 or more occurrences in the corpus), none of which is translated as a sign (the 6 occurrences for \"in\", the 4 for \"su\" and the 2 for \"di\" are due to alignment errors). We can conclude that prepositions do not exist in LIS as parts of speech; however, the prepositions in this cluster are often associated with morphological variations in the spatial positioning of related signs, which suggests that the role associated with these prepositions in Italian is performed by these variations in LIS. The conjunction \"e\" (and) also ends up in this cluster, although it has 8 legitimate sign alignments with \"pure\" (\"too\"); the rest are alignment errors. Unsurprisingly, all connectives in this class also have high ratings for the \"close\" hypothesis.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Subtree alignment and path processing",
"sec_num": "3.2"
},
{
"text": "We trained a classifier to help a LIS generator determine how an Italian connective should be translated. Because the translation pipeline we plan to integrate with is rule-based, we chose a Decision Tree as our classifier: this allows rules to be easily extracted from the classification model.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "In order to identify a single class for each example, we ranked the four possible translation candidates as follows: Sign is the strongest, then Location, then Close, and finally Missing is the child1 align = None \u2229 word = Per quanto riguarda \u2229 parent align = None \u21d2 Missing child1 align = None \u2229 word = Per quanto riguarda \u2229 parent align = PREVEDERE \u21d2 Close child1 align = None \u2229 word = o \u21d2 Align(O) child1 align = None \u2229 word = su \u2229 child2 align mykind = location \u2229 child2 align = SICILIA \u21d2 Location Figure 5 : Some rules extracted from the decision tree weakest. Then, each example is labeled with the strongest translation candidate available for it: thus, for example, if the connective word appears to be translated with a connective sign, and the words it connects are also aligned to signs which are close to each other syntactically, then the class is Sign, not Close.",
"cite_spans": [],
"ref_spans": [
{
"start": 502,
"end": 510,
"text": "Figure 5",
"ref_id": null
}
],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "Our training data suffers from large imbalance between the \"missing\" class and the others. A classifier that simply labels all examples as \"missing\" would have an accuracy above 60%, and in fact, that is the classifier that we obtain if we attempt to automatically optimize the parameters of a Decision Tree (DT). We also note that, for connectives where both options are possible, choosing to translate them can make the sentence more verbose, but choosing to drop them risks losing part of the sentence's meaning: the worse risk is the latter. Following accepted practice with unbalanced datasets (Chawla et al., 2004) , we rebalanced the classes by duplicating all examples of the Align, Location and Close classes, but not those of the Missing class.",
"cite_spans": [
{
"start": 599,
"end": 620,
"text": "(Chawla et al., 2004)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "On our data set of connectives with at least 10 occurrences, we trained a DT using AdaBoost (Freund and Schapire, 1997) . The features include the word neighboring the connective in the Italian syntax tree, their aligned signs if any, part of speech tags, and semantic categories such as time or location. The resulting tree is very large, but we provide a few examples of the rules that can be extracted from it in Figure 5 .",
"cite_spans": [
{
"start": 92,
"end": 119,
"text": "(Freund and Schapire, 1997)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [
{
"start": 416,
"end": 424,
"text": "Figure 5",
"ref_id": null
}
],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "Bootstrap evaluation shows our DT to have an accuracy of 83.58% \u00b1 1.03%. In contrast, a baseline approach of taking the most common class for each connective results in an accuracy of 68.70% \u00b1 0.88%. Furthermore, the baseline classifier has abysmal recall for the Close and Location classes (0.00% and 0.85%, respectively), which our DT greatly improves upon (86.73% and 75.32%).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "In order to estimate the impact of the lack of a LIS syntax tree in most of the corpus, we also learned and evaluated a DT using only the 1/4 of the corpus for which LIS syntax trees are available. The accuracy is 81.44% \u00b1 2.03%, versus a baseline of 71.55% \u00b1 1.74%. The recall for Close and Location is 89.22% and 73.58%, vs. 0.00% and 3.51% for the baseline. These results are comparable with the those obtained on the whole corpus, confirming that linear trees are a reasonable fallback.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "Both clustering and classification were performed using RapidMiner 5.3. 2",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule extraction",
"sec_num": "4"
},
{
"text": "The small size of our corpus, with around 375 bilingual sentences, posed a large challenge to the use of statistical methods; on the other hand, having no access to a LIS speaker prevented us from simply relying on a rule-based approach. By combining syntax tree processing with several machine learning techniques, we were able to analyze the corpus and detect patterns that show linguistic substance. We have produced initial results in terms of rule extraction, and we will be integrating these rules into the full Italian-LIS translation system to produce improved translation of connectives.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions and Future Work",
"sec_num": "5"
},
{
"text": "http://code.google.com/p/ berkeleyaligner/",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "This work was supported by the ATLAS project, funded by Regione Piemonte within the \"CIPE 2007\" framework. Partial support to the authors was also provided by awards IIS 0905593 (from the NSF) and NPRP 5-939-1-155 (from the QNRF). A special thanks to A. Mazzei (ATLAS) for his willingness to answer our email bursts. Thanks to other members of ATLAS, in particular P. Prinetto, N. Bertoldi, C. Geraci, L. Lesmo; and to C. Soria, who extracted the list of potential connectives from the Italian Treebank.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": "6"
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "A virtual character based italian sign language dictionary",
"authors": [
{
"first": "Nadeem",
"middle": [],
"last": "Ahmad",
"suffix": ""
},
{
"first": "Davide",
"middle": [],
"last": "Barberis",
"suffix": ""
},
{
"first": "Nicola",
"middle": [],
"last": "Garazzino",
"suffix": ""
},
{
"first": "Paolo",
"middle": [],
"last": "Prinetto",
"suffix": ""
},
{
"first": "Umar",
"middle": [],
"last": "Shoaib",
"suffix": ""
},
{
"first": "Gabriele",
"middle": [],
"last": "Tiotto",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Conference Universal Learning Design. Masaryk University",
"volume": "2",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Nadeem Ahmad, Davide Barberis, Nicola Garazzino, Paolo Prinetto, Umar Shoaib, and Gabriele Tiotto. 2012. A virtual character based italian sign language dictionary. In Proceedings of the Conference Uni- versal Learning Design. Masaryk University. 2 http://rapid-i.com/",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Arabic Text to Arabic Sign Language Translation System for the Deaf and Hearing-Impaired Community",
"authors": [
{
"first": "Abdulaziz",
"middle": [],
"last": "Almohimeed",
"suffix": ""
},
{
"first": "Mike",
"middle": [],
"last": "Wald",
"suffix": ""
},
{
"first": "R",
"middle": [
"I"
],
"last": "Damper",
"suffix": ""
}
],
"year": 2011,
"venue": "Proceedings of the Second Workshop on Speech and Language Processing for Assistive Technologies",
"volume": "",
"issue": "",
"pages": "101--109",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Abdulaziz Almohimeed, Mike Wald, and R.I. Damper. 2011. Arabic Text to Arabic Sign Language Trans- lation System for the Deaf and Hearing-Impaired Community. In Proceedings of the Second Work- shop on Speech and Language Processing for Assis- tive Technologies, pages 101-109, Edinburgh, Scot- land, UK, July. Association for Computational Lin- guistics.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Multilingual dependency parsing and domain adaptation using DeSR",
"authors": [
{
"first": "Giuseppe",
"middle": [],
"last": "Attardi",
"suffix": ""
},
{
"first": "Felice",
"middle": [],
"last": "Dell'orletta",
"suffix": ""
},
{
"first": "Maria",
"middle": [],
"last": "Simi",
"suffix": ""
},
{
"first": "Atanas",
"middle": [],
"last": "Chanev",
"suffix": ""
},
{
"first": "Massimiliano",
"middle": [],
"last": "Ciaramita",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the CoNLL Shared Task Session of EMNLP-CoNLL",
"volume": "",
"issue": "",
"pages": "1112--1118",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Giuseppe Attardi, Felice Dell'Orletta, Maria Simi, Atanas Chanev, and Massimiliano Ciaramita. 2007. Multilingual dependency parsing and domain adap- tation using DeSR. In Proceedings of the CoNLL Shared Task Session of EMNLP-CoNLL, pages 1112-1118.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "On the creation and the annotation of a large-scale Italian-LIS parallel corpus",
"authors": [
{
"first": "N",
"middle": [],
"last": "Bertoldi",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Tiotto",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Prinetto",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Piccolo",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Nunnari",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Lombardo",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Mazzei",
"suffix": ""
},
{
"first": "L",
"middle": [],
"last": "Damiano",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Lesmo",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Del Principe",
"suffix": ""
}
],
"year": 2010,
"venue": "4th Workshop on the Representation and Processing of Sign Languages: Corpora and Sign Language Technologies",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "N Bertoldi, G Tiotto, P Prinetto, E Piccolo, F Nunnari, V Lombardo, A Mazzei, R Damiano, L Lesmo, and A Del Principe. 2010. On the creation and the an- notation of a large-scale Italian-LIS parallel corpus. In 4th Workshop on the Representation and Process- ing of Sign Languages: Corpora and Sign Language Technologies, CSLT.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Sign language corpora for analysis, processing and evaluation",
"authors": [
{
"first": "Annelies",
"middle": [],
"last": "Braffort",
"suffix": ""
},
{
"first": "Laurence",
"middle": [],
"last": "Bolot",
"suffix": ""
},
{
"first": "Annick",
"middle": [],
"last": "Chtelat-Pel",
"suffix": ""
},
{
"first": "Maxime",
"middle": [],
"last": "Choisier",
"suffix": ""
},
{
"first": "Michael",
"middle": [],
"last": "Delorme",
"suffix": ""
},
{
"first": "J\u00e9r\u00e9mie",
"middle": [],
"last": "Filhol",
"suffix": ""
},
{
"first": "Cyril",
"middle": [],
"last": "Segouat",
"suffix": ""
},
{
"first": "Flora",
"middle": [],
"last": "Verrecchia",
"suffix": ""
},
{
"first": "Nad'ege",
"middle": [],
"last": "Badin",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Devos",
"suffix": ""
}
],
"year": 2010,
"venue": "Proc. of the Seventh International Conference on Language Resources and Evaluation (LREC'10)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Annelies Braffort, Laurence Bolot, E Chtelat-Pel, An- nick Choisier, Maxime Delorme, Michael Filhol, J\u00e9r\u00e9mie Segouat, Cyril Verrecchia, Flora Badin, and Nad'ege Devos. 2010. Sign language corpora for analysis, processing and evaluation. In Proc. of the Seventh International Conference on Language Re- sources and Evaluation (LREC'10).",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Editorial: special issue on learning from imbalanced data sets",
"authors": [
{
"first": "V",
"middle": [],
"last": "Nitesh",
"suffix": ""
},
{
"first": "Nathalie",
"middle": [],
"last": "Chawla",
"suffix": ""
},
{
"first": "Aleksander",
"middle": [],
"last": "Japkowicz",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Kotcz",
"suffix": ""
}
],
"year": 2004,
"venue": "ACM SIGKDD Explorations Newsletter",
"volume": "6",
"issue": "1",
"pages": "1--6",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Nitesh V Chawla, Nathalie Japkowicz, and Aleksander Kotcz. 2004. Editorial: special issue on learning from imbalanced data sets. ACM SIGKDD Explo- rations Newsletter, 6(1):1-6.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Dependency parsing with second-order feature maps and annotated semantic information",
"authors": [
{
"first": "Massimiliano",
"middle": [],
"last": "Ciaramita",
"suffix": ""
},
{
"first": "Giuseppe",
"middle": [],
"last": "Attardi",
"suffix": ""
}
],
"year": 2011,
"venue": "Trends in Parsing Technology",
"volume": "43",
"issue": "",
"pages": "87--104",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Massimiliano Ciaramita and Giuseppe Attardi. 2011. Dependency parsing with second-order feature maps and annotated semantic information. In H. Bunt, P. Merlo, and J. Nivre, editors, Trends in Parsing Technology, volume 43 of Text, Speech and Lan- guage Technology, pages 87-104. Springer.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Tailoring word alignments to syntactic machine translation",
"authors": [
{
"first": "John",
"middle": [],
"last": "Denero",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics (ACL-2007",
"volume": "",
"issue": "",
"pages": "17--24",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "John Denero. 2007. Tailoring word alignments to syn- tactic machine translation. In In Proceedings of the 45th Annual Meeting of the Association for Compu- tational Linguistics (ACL-2007, pages 17-24.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "SignSpeak-understanding, recognition, and translation of sign languages",
"authors": [
{
"first": "Philippe",
"middle": [],
"last": "Dreuw",
"suffix": ""
},
{
"first": "Jens",
"middle": [],
"last": "Forster",
"suffix": ""
},
{
"first": "Yannick",
"middle": [],
"last": "Gweth",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Stein",
"suffix": ""
},
{
"first": "Hermann",
"middle": [],
"last": "Ney",
"suffix": ""
},
{
"first": "Gregorio",
"middle": [],
"last": "Martinez",
"suffix": ""
},
{
"first": "Jaume",
"middle": [],
"last": "Verges Llahi",
"suffix": ""
},
{
"first": "Onno",
"middle": [],
"last": "Crasborn",
"suffix": ""
},
{
"first": "Ellen",
"middle": [],
"last": "Ormel",
"suffix": ""
},
{
"first": "Wei",
"middle": [],
"last": "Du",
"suffix": ""
}
],
"year": 2010,
"venue": "The 4th Workshop on the Representation and Processing of Sign Languages: Corpora and Sign Language Technologies (CSLT 2010)",
"volume": "",
"issue": "",
"pages": "22--23",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Philippe Dreuw, Jens Forster, Yannick Gweth, Daniel Stein, Hermann Ney, Gregorio Martinez, Jaume Verges Llahi, Onno Crasborn, Ellen Ormel, Wei Du, et al. 2010. SignSpeak-understanding, recognition, and translation of sign languages. In The 4th Workshop on the Representation and Pro- cessing of Sign Languages: Corpora and Sign Lan- guage Technologies (CSLT 2010), pages 22-23, Val- letta, Malta.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Learning non-isomorphic tree mappings for machine translation",
"authors": [
{
"first": "Jason",
"middle": [],
"last": "Eisner",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the 41st Annual Meeting on Association for Computational Linguistics",
"volume": "2",
"issue": "",
"pages": "205--208",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jason Eisner. 2003. Learning non-isomorphic tree mappings for machine translation. In Proceedings of the 41st Annual Meeting on Association for Compu- tational Linguistics-Volume 2, pages 205-208. As- sociation for Computational Linguistics.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Congiunzioni frasali, congiunzioni testuali e preposizioni: stessa logica, diversa testualit\u00e0",
"authors": [
{
"first": "Angela",
"middle": [],
"last": "Ferrari",
"suffix": ""
}
],
"year": 2008,
"venue": "",
"volume": "",
"issue": "",
"pages": "411--416",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Angela Ferrari. 2008. Congiunzioni frasali, congiun- zioni testuali e preposizioni: stessa logica, diversa testualit\u00e0. In Emanuela Cresti, editor, Prospettive nello studio del lessico italiano, Atti del IX Con- gresso della Societ\u00e0 Internazionale di Linguistica e Filologia Italiana, pages 411-416, Florence, Italy. Firenze University Press.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "A decision-theoretic generalization of on-line learning and an application to boosting",
"authors": [
{
"first": "Yoav",
"middle": [],
"last": "Freund",
"suffix": ""
},
{
"first": "Robert",
"middle": [
"E"
],
"last": "Schapire",
"suffix": ""
}
],
"year": 1997,
"venue": "Journal of computer and system sciences",
"volume": "55",
"issue": "1",
"pages": "119--139",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yoav Freund and Robert E Schapire. 1997. A decision-theoretic generalization of on-line learning and an application to boosting. Journal of computer and system sciences, 55(1):119-139.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "How grammar can cope with limited short-term memory: Simultaneity and seriality in sign languages",
"authors": [
{
"first": "Carlo",
"middle": [],
"last": "Geraci",
"suffix": ""
},
{
"first": "Marta",
"middle": [],
"last": "Gozzi",
"suffix": ""
},
{
"first": "Costanza",
"middle": [],
"last": "Papagno",
"suffix": ""
},
{
"first": "Carlo",
"middle": [],
"last": "Cecchetto",
"suffix": ""
}
],
"year": 2008,
"venue": "Cognition",
"volume": "106",
"issue": "2",
"pages": "780--804",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Carlo Geraci, Marta Gozzi, Costanza Papagno, and Carlo Cecchetto. 2008. How grammar can cope with limited short-term memory: Simultaneity and seriality in sign languages. Cognition, 106(2):780- 804.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Loosely tree-based alignment for machine translation",
"authors": [
{
"first": "Daniel",
"middle": [],
"last": "Gildea",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the 41st Annual Meeting on Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "80--87",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Daniel Gildea. 2003. Loosely tree-based alignment for machine translation. In Proceedings of the 41st Annual Meeting on Association for Computational Linguistics-Volume 1, pages 80-87. Association for Computational Linguistics.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Least squares quantization in pcm. Information Theory",
"authors": [
{
"first": "Stuart",
"middle": [],
"last": "Lloyd",
"suffix": ""
}
],
"year": 1982,
"venue": "IEEE Transactions on",
"volume": "28",
"issue": "2",
"pages": "129--137",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stuart Lloyd. 1982. Least squares quantization in pcm. Information Theory, IEEE Transactions on, 28(2):129-137.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "A virtual interpreter for the Italian Sign Language",
"authors": [
{
"first": "Vincenzo",
"middle": [],
"last": "Lombardo",
"suffix": ""
},
{
"first": "Fabrizio",
"middle": [],
"last": "Nunnari",
"suffix": ""
},
{
"first": "Rossana",
"middle": [],
"last": "Damiano",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the 10th International Conference on Intelligent Virtual Agents, IVA'10",
"volume": "",
"issue": "",
"pages": "201--207",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Vincenzo Lombardo, Fabrizio Nunnari, and Rossana Damiano. 2010. A virtual interpreter for the Italian Sign Language. In Proceedings of the 10th International Conference on Intelligent Virtual Agents, IVA'10, pages 201-207, Berlin, Heidelberg. Springer-Verlag.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "An avatarbased interface for the Italian Sign Language",
"authors": [
{
"first": "Vincenzo",
"middle": [],
"last": "Lombardo",
"suffix": ""
},
{
"first": "Cristina",
"middle": [],
"last": "Battaglino",
"suffix": ""
},
{
"first": "Rossana",
"middle": [],
"last": "Damiano",
"suffix": ""
},
{
"first": "Fabrizio",
"middle": [],
"last": "Nunnari",
"suffix": ""
}
],
"year": 2011,
"venue": "2011 International Conference on Complex, Intelligent and Software Intensive Systems (CISIS)",
"volume": "",
"issue": "",
"pages": "589--594",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Vincenzo Lombardo, Cristina Battaglino, Rossana Damiano, and Fabrizio Nunnari. 2011. An avatar- based interface for the Italian Sign Language. In 2011 International Conference on Complex, Intelli- gent and Software Intensive Systems (CISIS), pages 589-594. IEEE.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Source Language Categorization for improving a Speech into Sign Language Translation System",
"authors": [
{
"first": "Rub\u00e9n",
"middle": [],
"last": "Ver\u00f3nica L\u00f3pez-Lude\u00f1a",
"suffix": ""
},
{
"first": "Syaheerah",
"middle": [],
"last": "San-Segundo",
"suffix": ""
},
{
"first": "Juan",
"middle": [
"Manuel"
],
"last": "Lufti",
"suffix": ""
},
{
"first": "Juli\u00e1n",
"middle": [],
"last": "Lucas-Cuesta",
"suffix": ""
},
{
"first": "Beatriz",
"middle": [],
"last": "David Echevarry",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Mart\u00ednez-Gonz\u00e1lez",
"suffix": ""
}
],
"year": 2011,
"venue": "Proceedings of the Second Workshop on Speech and Language Processing for Assistive Technologies",
"volume": "",
"issue": "",
"pages": "84--93",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ver\u00f3nica L\u00f3pez-Lude\u00f1a, Rub\u00e9n San-Segundo, Syaheerah Lufti, Juan Manuel Lucas-Cuesta, Juli\u00e1n David Echevarry, and Beatriz Mart\u00ednez- Gonz\u00e1lez. 2011. Source Language Categorization for improving a Speech into Sign Language Trans- lation System. In Proceedings of the Second Workshop on Speech and Language Processing for Assistive Technologies, pages 84-93, Edinburgh, Scotland, UK, July.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Automatic categorization for improving Spanish into Spanish Sign Language machine translation",
"authors": [
{
"first": "Juan",
"middle": [
"Manuel"
],
"last": "Montero",
"suffix": ""
},
{
"first": "Ricardo",
"middle": [],
"last": "C\u00f3rdoba",
"suffix": ""
},
{
"first": "Javier",
"middle": [],
"last": "Ferreiros",
"suffix": ""
},
{
"first": "Jos\u00e9 Manuel",
"middle": [],
"last": "Pardo",
"suffix": ""
}
],
"year": 2011,
"venue": "Computer Speech & Language",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Juan Manuel Montero, Ricardo C\u00f3rdoba, Javier Ferreiros, and Jos\u00e9 Manuel Pardo. 2011. Automatic categorization for improving Spanish into Spanish Sign Language machine translation. Computer Speech & Language.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Collecting a Motion-Capture Corpus of American Sign Language for Data-Driven Generation Research",
"authors": [
{
"first": "Pengfei",
"middle": [],
"last": "Lu",
"suffix": ""
},
{
"first": "Matt",
"middle": [],
"last": "Huenerfauth",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the NAACL HLT 2010 Workshop on Speech and Language Processing for Assistive Technologies",
"volume": "",
"issue": "",
"pages": "89--97",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pengfei Lu and Matt Huenerfauth. 2010. Collecting a Motion-Capture Corpus of American Sign Lan- guage for Data-Driven Generation Research. In Proceedings of the NAACL HLT 2010 Workshop on Speech and Language Processing for Assistive Technologies, pages 89-97, Los Angeles, California, June. Association for Computational Linguistics.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Learning a Vector-Based Model of American Sign Language Inflecting Verbs from Motion-Capture Data",
"authors": [
{
"first": "Pengfei",
"middle": [],
"last": "Lu",
"suffix": ""
},
{
"first": "Matt",
"middle": [],
"last": "Huenerfauth",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Third Workshop on Speech and Language Processing for Assistive Technologies",
"volume": "",
"issue": "",
"pages": "66--74",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pengfei Lu and Matt Huenerfauth. 2012. Learn- ing a Vector-Based Model of American Sign Lan- guage Inflecting Verbs from Motion-Capture Data. In Proceedings of the Third Workshop on Speech and Language Processing for Assistive Technolo- gies, pages 66-74, Montr\u00e9al, Canada, June. Asso- ciation for Computational Linguistics.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Some methods for classification and analysis of multivariate observations",
"authors": [
{
"first": "James",
"middle": [],
"last": "Macqueen",
"suffix": ""
}
],
"year": 1967,
"venue": "Proceedings of the fifth Berkeley symposium on mathematical statistics and probability",
"volume": "1",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "James MacQueen. 1967. Some methods for classi- fication and analysis of multivariate observations. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability, volume 1, page 14. California, USA.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "Syntactic realignment models for machine translation",
"authors": [
{
"first": "Jonathan",
"middle": [],
"last": "May",
"suffix": ""
},
{
"first": "Kevin",
"middle": [],
"last": "Knight",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of EMNLP",
"volume": "",
"issue": "",
"pages": "360--368",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jonathan May and Kevin Knight. 2007. Syntactic re- alignment models for machine translation. In Pro- ceedings of EMNLP, pages 360-368.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "Sign language generation with expert systems and ccg",
"authors": [
{
"first": "Alessandro",
"middle": [],
"last": "Mazzei",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the Seventh International Natural Language Generation Conference, INLG '12",
"volume": "",
"issue": "",
"pages": "105--109",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alessandro Mazzei. 2012. Sign language generation with expert systems and ccg. In Proceedings of the Seventh International Natural Language Generation Conference, INLG '12, pages 105-109, Strouds- burg, PA, USA. Association for Computational Lin- guistics.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Building the Italian syntactic-semantic Treebank",
"authors": [
{
"first": "Simonetta",
"middle": [],
"last": "Montemagni",
"suffix": ""
},
{
"first": "Francesco",
"middle": [],
"last": "Barsotti",
"suffix": ""
},
{
"first": "Marco",
"middle": [],
"last": "Battista",
"suffix": ""
},
{
"first": "Nicoletta",
"middle": [],
"last": "Calzolari",
"suffix": ""
},
{
"first": "Ornella",
"middle": [],
"last": "Corazzari",
"suffix": ""
},
{
"first": "Alessandro",
"middle": [],
"last": "Lenci",
"suffix": ""
},
{
"first": "Antonio",
"middle": [],
"last": "Zampolli",
"suffix": ""
},
{
"first": "Francesca",
"middle": [],
"last": "Fanciulli",
"suffix": ""
},
{
"first": "Maria",
"middle": [],
"last": "Massetani",
"suffix": ""
},
{
"first": "Remo",
"middle": [],
"last": "Raffaelli",
"suffix": ""
}
],
"year": 2003,
"venue": "Treebanks",
"volume": "",
"issue": "",
"pages": "189--210",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Simonetta Montemagni, Francesco Barsotti, Marco Battista, Nicoletta Calzolari, Ornella Corazzari, Alessandro Lenci, Antonio Zampolli, Francesca Fanciulli, Maria Massetani, Remo Raffaelli, et al. 2003. Building the Italian syntactic-semantic Tree- bank. Treebanks, pages 189-210.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Building Sign Language Corpora for Use in Machine Translation",
"authors": [
{
"first": "S",
"middle": [],
"last": "Morrissey",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Somers",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Smith",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Gilchrist",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Dandapat",
"suffix": ""
}
],
"year": 2010,
"venue": "4th Workshop on the Representation and Processing of Sign Languages: Corpora and Sign Language Technologies (CSLT 2010)",
"volume": "",
"issue": "",
"pages": "172--178",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Morrissey, H. Somers, R. Smith, S. Gilchrist, and S. Dandapat. 2010. Building Sign Language Cor- pora for Use in Machine Translation. In 4th Work- shop on the Representation and Processing of Sign Languages: Corpora and Sign Language Technolo- gies (CSLT 2010), pages 172-178, Valletta, Malta.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "The Italian Sign Language Sign Bank: Using WordNet for Sign Language corpus creation",
"authors": [
{
"first": "P",
"middle": [],
"last": "Prinetto",
"suffix": ""
},
{
"first": "U",
"middle": [],
"last": "Shoaib",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Tiotto",
"suffix": ""
}
],
"year": 2011,
"venue": "2011 International Conference on Communications and Information Technology (ICCIT)",
"volume": "",
"issue": "",
"pages": "134--137",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "P. Prinetto, U. Shoaib, and G. Tiotto. 2011. The Ital- ian Sign Language Sign Bank: Using WordNet for Sign Language corpus creation. In 2011 Interna- tional Conference on Communications and Informa- tion Technology (ICCIT), pages 134-137.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "Dizionario dei segni: la lingua dei segni in 1400 immagini",
"authors": [
{
"first": "Orazio",
"middle": [],
"last": "Romeo",
"suffix": ""
}
],
"year": 1991,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Orazio Romeo. 1991. Dizionario dei segni: la lingua dei segni in 1400 immagini. Zanichelli.",
"links": null
},
"BIBREF29": {
"ref_id": "b29",
"title": "La lingua italiana dei segni: la comunicazione visivo-gestuale dei sordi",
"authors": [
{
"first": "Virginia",
"middle": [],
"last": "Volterra",
"suffix": ""
}
],
"year": 1987,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Virginia Volterra. 1987. La lingua italiana dei segni: la comunicazione visivo-gestuale dei sordi. Il Mulino.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"uris": null,
"num": null,
"type_str": "figure",
"text": "Example excerpt from an AEWLIS file"
},
"FIGREF1": {
"uris": null,
"num": null,
"type_str": "figure",
"text": "Ratio of ITA/LIS occurrences in logarithmic scale."
},
"FIGREF2": {
"uris": null,
"num": null,
"type_str": "figure",
"text": "19_f08_2011-06-08_10_04_10.xml Anche sulla Sardegna qualche annuvolamento pomeridiano, possibilit\u00e0 di qualche breve scroscio di pioggia, ma tendenza poi a schiarite pomeriggio Sardegna area nuvola pure acquazzone potere ma poi nuvola Example of integrated syntax trees."
},
"TABREF0": {
"num": null,
"type_str": "table",
"html": null,
"text": "Sardinia skies will become overcast in the afternoon, chance of a few brief downpours of rain, but then a trend towards a mix of sun and clouds\".",
"content": "<table><tr><td colspan=\"2\">(1) (Ita.) Anche Also</td><td>sulla on</td><td colspan=\"3\">Sardegna Sardinia</td><td>qualche a few</td><td colspan=\"2\">annuvolamento cloud covers</td><td colspan=\"3\">pomeridiano, afternoon[adj],</td><td>possibilit\u00e0 chance</td><td>di of</td><td>qualche a few</td><td>breve brief</td><td>scroscio downpour</td><td>di of</td><td>pioggia, rain,</td></tr><tr><td>ma but</td><td>tendenza trend</td><td>poi then</td><td colspan=\"2\">a towards</td><td colspan=\"3\">schiarite. sunny spells.</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"4\">\"Also on (2) (LIS) POMERIGGIO Afternoon</td><td colspan=\"3\">SARDEGNA Sardinia</td><td>AREA area</td><td colspan=\"2\">NUVOLA cloud</td><td>PURE also</td><td colspan=\"2\">ACQUAZZONE downpour</td><td>POTERE can[modal]</td><td>MA but</td><td>POI then</td><td>NUVOLA cloud</td></tr><tr><td colspan=\"3\">DIMINUIRE decrease</td><td/><td/><td/><td/><td/><td/><td/><td/><td/><td/></tr></table>"
},
"TABREF2": {
"num": null,
"type_str": "table",
"html": null,
"text": "Translation candidates for connectives with more than 10 occurrences",
"content": "<table><tr><td>Connective</td><td>ITA Occurrences Sign</td><td>Location</td><td>Close</td><td>Missing</td></tr><tr><td>domani dopodomani mentre o per\u00f2 ancora invece ma poi</td><td>71 67 (94.37%) 15 14 (93.33%) 28 26 (92.86%) 37 37 (100.00%) 10 9 (90.00%) 72 44 (61.11%) 17 9 (52.94%) 51 29 (56.86%)</td><td>1 (1.41%) 0 (0.00%) 5 (17.86%) 2 (5.41%) 1 (10.00%) 1 (1.39%) 1 (5.88%) 1 (1.96%)</td><td>2 (2.82%) 0 (0.00%) 0 (0.00%) 6 (16.22%) 1 (10.00%) 3 (4.17%) 2 (11.76%) 2 (3.92%)</td><td>4 (5.63%) 1 (6.67%) 1 (3.57%) 0 (0.00%) 1 (10.00%) 25 (34.72%) 6 (35.29%) 21 (41.18%)</td></tr></table>"
}
}
}
} |