{"id":1040,"date":"2015-07-20T17:09:31","date_gmt":"2015-07-20T17:09:31","guid":{"rendered":"http:\/\/blogs.dickinson.edu\/dcc\/?p=1040"},"modified":"2015-12-31T12:14:24","modified_gmt":"2015-12-31T12:14:24","slug":"johan-winges-new-latin-macronizer","status":"publish","type":"post","link":"https:\/\/blogs.dickinson.edu\/dcc\/2015\/07\/20\/johan-winges-new-latin-macronizer\/","title":{"rendered":"Johan Winge’s New Latin Macronizer"},"content":{"rendered":"
\"Inscription_latine_avec_apex_extrait\"<\/a>

image credit: Vincent Ramos via Wikimedia Commons<\/p><\/div>\n

A new Latin macronizer<\/a> has come on the scene, and it is superb. It should become an essential tool for Latin teachers and editors of Latin texts. The author is Johan Winge, who just completed his undergraduate studies in the Language Technology Programme at Uppsala University, supervised by Joakim Nivre. The macronizer is the result of his thesis work for the degree. I had the opportunity to give it a good test run recently, as I read the Ilias Latina\u00a0<\/em>along with about twenty Latin teachers at the Dickinson Summer Latin Workshop. I took the PHI text (Vollmer’s Teubner from 1913) of this 1070-line condensation of the Iliad<\/em> into Latin hexameters, put it in a Word document, and ran it through Winge’s macronizer. We read the text together and spotted the cases where corrections were needed.<\/p>\n

The claim on the site that “The expected accuracy on an average classical text is estimated to be about 98% to 99%” seems like no exaggeration. What makes Winge’s macronizer more effective that other\u00a0tools\u00a0such as Kevin Ryan’s Macron Helper<\/a> or Felipe Vogel’s m\u0101ccer<\/a>\u00a0is that it does not work on the basis of a database of previously macronized forms. Rather, it uses\u00a0a part-of-speech tagger (RFTagger<\/a>) trained on the Latin Dependency Treebank<\/a>, and with macrons provided by a customized version of the Morpheus morphological analyzer.<\/p>\n

You’ll have to read Johan’s thesis, Automatic Annotation of Latin Vowel Length<\/a>, to get all the technical details. I’ll just say that it performed splendidly on the Ilias Latina<\/em>. Here is a typical stretch, lines 344-374, with the errors highlighted:<\/p>\n

dumque inter s\u0113s\u0113 procer\u0113s cert\u0101men hab\u0113rent,
\nconcilium omnipot\u0113ns habuit r\u0113gn\u0101tor Olymp\u012b 345
\nfoederaque intent\u014d turb\u0101vit Pandarus arc\u016b,
\nt\u0113, Menel\u0101e, pet\u0113ns; lat\u0113<\/span>r\u012bque vol\u0101tile t\u0113lum
\ninc\u012b<\/span>dit et tunicam ferr\u014d squ\u0101m\u012bsque rigentem
\ndissecat: exc\u0113dit pugna<\/span> gemebundus Atr\u012bd\u0113s
\ncastraque t\u016bta petit; quem doctus ab arte patern\u0101 350
\nPae\u014dni\u012bs c\u016brat iuvenis Podal\u012brius herb\u012bs
\nitque iterum in caed\u0113s horrendaque proelia victor.
\narm\u0101vit fort\u0113s Agamemnonis \u012bra Pelasg\u014ds
\net dolor in pugnam c\u016bnct\u014ds comm\u016bn\u012b<\/span>s ag\u0113bat.
\nbellum ing\u0113ns oritur multumque utrimque cru\u014dris 355
\nfunditur et t\u014dt\u012bs sternuntur corpora camp\u012bs;
\ninque vicem Tr\u014dumque cadunt Danaumque catervae.
\nnec requi\u0113s datur \u016blla vir\u012bs; sonat undique Ma<\/span>vors
\nt\u0113l\u014drumque volant c\u016bnct\u012bs \u0113 partibus imbr\u0113s.
\nocc\u012b<\/span>dit Antiloch\u012b rigid\u014d d\u0113mersus in umbr\u0101s 360
\n\u0113nse Thalysiad\u0113s opt\u0101taque l\u016bmina linquit.
\ninde man\u016b fort\u012b Gr\u0101i\u014drum terga prementem
\noccupat Anthemi\u014dne satum Telam\u014dnius Ai\u0101x
\net praed\u016br\u0101t\u014d tr\u0101nsf\u012bxit pectora t\u0113l\u014d:
\npurpure\u014d vomit ille animam cum sanguine mixtam, 365
\n\u014dra rigat mori\u0113ns. tum magn\u012bs Antiphus hastam
\nv\u012bribus adversum c\u014dn\u0101t\u016b<\/span>s corpore t\u014dt\u014d
\ntorquet in Aeacid\u0113n: t\u0113lumque err\u0101vit ab hoste
\ninque hostem cecidit, tr\u0101nsf\u012bxit et inguine Leuc\u014dn:
\nconc\u012b<\/span>dit \u012bnf\u0113l\u012bx pr\u014dstr\u0101tus vulnere fort\u012b 370
\net carpit virid\u0113s moribundus dentibus herb\u0101s.
\n\u2020impiger \u2020Atr\u012bd\u0113s c\u0101s\u016b concuss\u016b<\/span>s am\u012bc\u012b
\nDemocoonta petit t\u0113l\u014dque advers\u0101<\/span> trab\u0101l\u012b
\ntempora\u00a0tr\u0101nsadigit …<\/p><\/blockquote>\n

You will note that of the 11 “mistakes” on this page, only one (Mavors<\/em>) is a genuine error. All the others are simply ambiguous forms, issues that need to be decided\u00a0by a human. Virtually all of the cases\u00a0that did not fall into the category of “ambiguous forms that need to be decided by a human” were Greek proper names, in which this text abounds. For some reason the form\u00a0Achillis<\/em> consistently came out with a long mark on the final vowel. Paris came out with a final macron twice, but without it three times. There were quantity issues with Nereus, and his daughters.The strange form m\u0113\u014d emerged at line 851.\u00a0But virtually all the time, with all ordinary Latin words, the macronizer performed brilliantly. The greatest delight was seeing it correctly macronize the phrase\u00a0r\u0113bus in art\u012bs\u00a0<\/em>(line 968), where the final word almost always has a short “i”–but not here. That will have been\u00a0the result of the Treebank data, I am guessing.<\/p>\n

Mr. Winge, I salute you!<\/p>\n

Postrcipt 7\/21\/15<\/strong>: Johan writes that his source code<\/a> is now available. \u00a0Also,\u00a0the picture I posted originally is not of him but of his friend Francesco Veneziano. Apologies to both Johan and Francesco for that one!<\/p>\n","protected":false},"excerpt":{"rendered":"

A new Latin macronizer has come on the scene, and it is superb. It should become an essential tool for Latin teachers and editors of Latin texts. The author is Johan Winge, who just completed his undergraduate studies in the … Continue reading →<\/span><\/a><\/p>\n","protected":false},"author":65,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ngg_post_thumbnail":0,"footnotes":""},"categories":[1],"tags":[95797,95793],"class_list":["post-1040","post","type-post","status-publish","format-standard","hentry","category-uncategorized","tag-johan-winge","tag-macronizers"],"_links":{"self":[{"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/posts\/1040","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/users\/65"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/comments?post=1040"}],"version-history":[{"count":0,"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/posts\/1040\/revisions"}],"wp:attachment":[{"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/media?parent=1040"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/categories?post=1040"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.dickinson.edu\/dcc\/wp-json\/wp\/v2\/tags?post=1040"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}