{"id":938,"date":"2020-06-09T16:31:32","date_gmt":"2020-06-09T16:31:32","guid":{"rendered":"http:\/\/www.wellformedness.com\/blog\/?p=938"},"modified":"2020-06-23T14:50:11","modified_gmt":"2020-06-23T14:50:11","slug":"results-sigmorphon-2020-shared-task-multilingual-grapheme-phoneme-conversion","status":"publish","type":"post","link":"https:\/\/www.wellformedness.com\/blog\/results-sigmorphon-2020-shared-task-multilingual-grapheme-phoneme-conversion\/","title":{"rendered":"Results of the SIGMORPHON 2020 shared task on multilingual grapheme-to-phoneme conversion"},"content":{"rendered":"<p>The results of the <a href=\"https:\/\/sigmorphon.github.io\/sharedtasks\/2020\/task1\/\">SIGMORPHON 2020 shared task on multilingual grapheme-to-phoneme conversion<\/a> are now in, and are summarized in <a href=\"https:\/\/www.aclweb.org\/anthology\/2020.sigmorphon-1.2\/\">our task paper<\/a>. A couple bullet points:<\/p>\n<ul>\n<li>Unsurprisingly, the best systems all used some form of ensembling.<\/li>\n<li>Many of the best teams performed self-training and\/or data augmentation experiments, but most of these experiments were performance-negative except in simulated low-resource conditions. Maybe we&#8217;ll do a low-resource challenge in a future year.<\/li>\n<li>LSTMs and transformers are roughly neck-and-neck; one strong submission used a variant of <a href=\"https:\/\/www.aclweb.org\/anthology\/P17-1183\/\">hard monotonic attention<\/a>.<\/li>\n<li>Many of the best teams used some kind of pre-processing romanization strategy for Korean, the language with the worst baseline accuracy. We speculate why this helps in the task paper.<\/li>\n<li>There were some concerns about data quality for three languages (Bulgarian, Georgian, and Lithuanian). <a href=\"https:\/\/en.wiktionary.org\/wiki\/Wiktionary:Information_desk\/2020\/April#Performing_bulk_edits\">We know how to fix them<\/a> and will do so this summer, if time allows. We may also &#8220;re-issue&#8221; the challenge data with these fixes.<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>The results of the SIGMORPHON 2020 shared task on multilingual grapheme-to-phoneme conversion are now in, and are summarized in our task paper. A couple bullet points: Unsurprisingly, the best systems all used some form of ensembling. Many of the best teams performed self-training and\/or data augmentation experiments, but most of these experiments were performance-negative except &hellip; <a href=\"https:\/\/www.wellformedness.com\/blog\/results-sigmorphon-2020-shared-task-multilingual-grapheme-phoneme-conversion\/\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;Results of the SIGMORPHON 2020 shared task on multilingual grapheme-to-phoneme conversion&#8221;<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"aside","meta":{"_crdt_document":"","footnotes":""},"categories":[3,4,5,21],"tags":[],"class_list":["post-938","post","type-post","status-publish","format-aside","hentry","category-dev","category-language","category-nlp","category-speech","post_format-post-format-aside"],"_links":{"self":[{"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/posts\/938","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/comments?post=938"}],"version-history":[{"count":2,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/posts\/938\/revisions"}],"predecessor-version":[{"id":943,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/posts\/938\/revisions\/943"}],"wp:attachment":[{"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/media?parent=938"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/categories?post=938"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wellformedness.com\/blog\/wp-json\/wp\/v2\/tags?post=938"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}