Page MenuHomePhabricator

CX2: Issues with translating articles with <score> tags
Open, MediumPublic

Description

Split form T139457: Support <score> tags in Content Translation.

  1. Start translating an article that includes <score> tags - e.g. en: Also sprach Zarathustra (Strauss) or en: An Alpine Symphony to any language.
  2. Several issues are present
    • the articles cannot be translated paragraph by paragraph - when you click in the Translation panel to start translation, the article will be translated in one piece:

Screen Shot 2019-02-21 at 3.24.23 PM.png (723×1 px, 229 KB)

  • the paragraphs are misaligned

Screen Shot 2019-02-22 at 4.29.09 PM.png (791×1 px, 417 KB)

  • MT options silently fail

However, the translation successfully displays musical notations.

Event Timeline

I just want to add some general notes here, as we work out what is going on in this page and in other related pages. (The same issue can be found in ALL of the pages listed under the category "Tone Poems by Richard Strauss.")

Last week I did a lot of testing of the <score> tag in CX2, on pages OTHER than "Also Sprach..." and its cousins, and I found almost no problems. (A couple that I tested: "Jazz" and "ABC notation." Also see "Symphony No. 9 (Beethoven)" )
Here are the results I posted:

Results of testing of <score> tag support

  • Tested on these language pairs: en>iw, en>zh, en>gu, en>es, en>ji, fr>ru
  • Tested both Lilypond and ABC notation.
  • Tested both block and inline <score> elements.
  • Tested all translation engines available for the language pair.

Could find only ONE instance in which a <score> element was not translated, and that was when testing en>zh and using the Youdao translation engine. I could successfully translate plain text with Youdao, but when I attempted a <score> element, I got the error message "machine translation failed," and the translation block was left empty.

In every other situation, translation of <score> worked perfectly. Even in the specific page and language pair mentioned above
(https://en.wikipedia.org/wiki/Also_sprach_Zarathustra_(Strauss) in en>iw), <score> elements translated without trouble, even
though the bulk of the normal text did not (because apparently the source article uses templates that aren't available in the
target language.)

The fact that the <score> tag seems to translate without problems on other pages makes me wonder if there is something OTHER than the <score> tag that is causing "Also Sprach" and the related pages in the "Tone Poems of Richard Strauss" category to translate as a single blob.

Just posing the question. I've been looking at the wikitext in the "Tone Poem" pages, searching for some feature that's common to them all but not to the many other pages I've tested, and so far I've come up empty-handed.

A pattern emerges (that doesn’t have to do with <score>)...

ALL of the articles in the category "Tone Poems by Richard Strauss" share the same problem as Also Sprach Zarathustra: They translate as a single block or blob rather than as individual paragraphs.

One trait that all of them share is that they start with:

{{italic title}}
[[File:Strauss1894.jpg|thumb|upright|Richard Strauss, 1894]]

I checked other articles that begin with {{italic title}}. By itself, it does not seem to cause articles to be translated as a single blob. For example, see: “Boston Herald”, “Dr. Who”, “Carousel”, “Götterdämmerung”, “An American in Paris” “Pteranodon”.

However, every article I found that, like “Also Sprach”and its cousins, started with {{italic title}} followed by a captioned thumbnail ALSO translated as a single blob. Here are some examples. (The title is followed by the wikitext that begins the article.)

Eine Kleine Nachtmusik

{{Italic title}}
[[File:Wolfgang-amadeus-mozart 1.jpg|thumb|upright|Posthumous portrait of Mozart (1819,&nbsp;[[Barbara Krafft]])]]

Héroïde funèbre (Liszt)

{{italic title}}
[[File:Battle at Tápióbicske.jpg|thumb|A portrayal of the Battle at Tápióbicske, a battle between Hungarian revolutionists and the Hungarian Government]]

Les préludes

{{italic title}}
[[Image:Liszt-kaulbach.jpg|right|thumb|Franz Liszt, after a painting of 1856, by [[Wilhelm von Kaulbach]].]]

Wakizashi

{{Italic title|reason=[[:Category:Japanese words and phrases]]}}
[[File:Daisho katana and wakizashi 1.jpg|thumb|350px|A Daito (top) and wakizashi (bottom) in the form of a ''[[Daisho|daishō]]'', showing the difference in size.]]

Note that the same does not happen when {{DISPLAY TITLE:…}} is used instead of {{italic title}}. Example:

“Themes from Mozart’s Figaro and Don Giovanni”

{{DISPLAYTITLE:''Fantasy on Themes from Mozart's ''Figaro'' and ''Don Giovanni}}
[[File:Cover of 'Fantasie über Themen aus Mozarts Figaro und Don Giovanni' v2.jpg|thumb|Cover page of the 1997 score]]

A few stray notes:

  • {{italic title}} seems to be ignored or somehow stripped out in CX2. Though the Wikipedia article has an italicized title, in CX2 the title is not italicized (in either the source or translation column), and it is also not italicized in the published translation. (The same seems to be true for {{DISPLAYTITLE}}.
  • The pattern discussed here is for articles whose wikitext STARTS with with {{italic title}} and is followed by a thumbnail image. Those articles seem to always translate as a blob. But I want to mention two examples that almost matched the pattern, but in each case the {{Italic image}} template was preceded by another template.

In one case, the page translated normally:
A Hero's Song

{{good article}}
{{italic title}}
[[File:Hero's Song.jpg|thumb|Title page of the first edition of ''A Hero's Song'', published in 1899.]]

In the other, the page still translated as a blob:
Theogony

{{short description|Poem by Hesiod}}
{{italic title}}
[[File:Hesiod, Theogony, Venice, Gr. 464.jpg|thumb|Fourteenth-century [[Greek language|Greek]]

Thx, @Barbvd!
The title styling recommendation can be found in https://en.wikipedia.org/wiki/Wikipedia:Article_titles#Article_title_format. So, the fact that ContentTranslation does not preserve the title style might negatively affect the review of the article.

Arrbee renamed this task from Issues with translating articles with <score> tags to CX2: Issues with translating articles with <score> tags.Mar 4 2019, 2:51 PM
Arrbee triaged this task as Medium priority.
Arrbee moved this task from Needs Triage to Bugs on the ContentTranslation board.