MITRE CLASR Project Spanish Broadcast Translation Guidelines
Version 1.4 May 6, 2005
NOTE: These guidelines were also used for the Mandarin translations. Since the translation house had already used them for Spanish, they were able to apply them to the Mandarin data with no changes.
1. Goal
The goal of this project is to translate a set of transcribed radio broadcasts from Spanish to English. We want to make sure primarily that all of the information in the Spanish is present in the English version. Secondarily, the English should be readable.
2. Document Format
The files will have the following format, derived from the original transcripts of Spanish language audio broadcasts. Example:
Líder del cartel del Golfo rinde su declaración preparatoria. Y a continuación, ya la vieron ustedes aquí está Leti Vivanco, que nos lleva por otros rumbos de la noticia. Buenos días, Leti.
Please translate segment-by-segment the text next to the tags into the currently empty area indicated by the tags. Here's where the translations would go in the previous example:
Líder del cartel del Golfo rinde su declaración preparatoria.
Y a continuación, ya la vieron ustedes aquí está Leti Vivanco, que nos lleva por otros rumbos de la noticia. Place English version of segment marked with here Buenos días, Leti. Place English version of segment marked with here
Translate the segments in the order given. Note that information about the speaker is given in the tag. There are other tags throughout the file that should be left untranslated. Some of them (e.g. ) may be informative for your translaitons. Please do not modify or delete any text inside of the tag boundaries (between < and >).
3. Translation Style
The English translation must be faithful to the original Spanish text in terms of meaning and style. The Spanish source text is usually a news broadcast, thus the translation should read like a news broadcast. Try to translate as literally as possible, but without sacrificing fluency or naturalness. In sections where the radio broadcasts are conversational, conversational style should be used in the translations. Most of the disfluencies from the original broadcasts were removed in the transcriptions. To the extent possible without sacrificing meaning, please remove the remainder during translation. Because the source text is only a partially edited transcription of the spoken broadcast, it might be hard to read in places. It might make more sense to read it aloud in those places. We have tried to clean up awkward-sounding hesitations and mispronunciations, but there may be places where they remain or where the resulting text is not very clean. All of this text was spoken this way in the broadcasts.
3.1. Spell-checking, translation aides
Please try to keep spelling consistent throughout your translations and perform spell-checking if possible. If you use automated spell checking software, please let us know which one you use. Also, please let us know which translation reference aides you use, such as translation memories or machine translations.
3.2. Numeric references
Numeric entities, such as dates, phone numbers, monetary amounts, population sizes, etc., can vary in how they are presented, either as digit strings or as words. (examples: “a million”, “1,000,000”, “1 000 000”, “1000000”, “1e+06” ; “24 June 2005”, “June 24, 2005”, “6/24/2005”, “24/6/2005” ; “four hundred” , “400”)
As you translate numeric references like these, please write them the way you would expect to see them in English transcripts. Try to be consistent with the rest of the translators.
3.3. Proper names
Proper names (people, places, organizations) can sometimes be hard to translate. If no ready translation springs to mind, then transliterate the name into a form you'd expect other English readers would accept and put the original word in brackets [] just after the transliteration. Please do your best to translate well-known person and place names as they are normally written in English. Proper names are not expected to be a big problem for Spanish to English translation.\
4. Encoding
The source document is encoded in Latin-1, aka ISO-8859-1 format. The translation results should be either 1) kept in Latin-1, or 2) produced in unicode, UTF-8. Please make all files consistent, though, and let us know the final encoding.
5. Addendum on translation style
At a early point in the Spanish translation process, our translation agency asked for a clarification on how to translate style, presenting examples (A) and (B) below. The more literal and conversationa (B) was chosen as superior and more appropriate for our needs. No obstante, el estado está garantizando la protección de salud en instituciones financiadas por el estado. También, igual, de igual modo, la escolaridad, las escuelas la educación es obligatoria hasta la edad de niños de, hasta dieciocho años y, en todas las escuelas públicas, todas las escuelas públicas son gratuitas, aunque la constitución, nueva constitución permite que algunas, algunos elementos del sistema educacional público sean pagaderos. This could be translated as (A) or (B): (A) However, the State is guaranteeing health protection in State-funded institutions. And schooling also works the same way: school education is compulsory until the age of 18, and all public schools are free, although the new Constitution allows some State schools to charge fees. (B) However, the State is guaranteeing health protection in State-funded institutions. Also, equally, in the same way, schooling. School education is compulsory for children of up to 18 years of age, and in all State schools, all State
schools are free, although the Constitution, the new Constitution allows some of them, some elements of the State education system to charge fees.