Anagram Translation Guidelines PDF

Anagram Translation Guidelines Project Instructions Table of Contents Introduction..........................................................................................................................................

Anagram Translation Guidelines Project Instructions Table of Contents Introduction...................................................................................................................................................................................... 1 Project Overview.............................................................................................................................................................................. 2 Translation Guidelines......................................................................................................................................................................... 2 A. Terminology.......................................................................................................................................................................... 2 B. Assumptions......................................................................................................................................................................... 2 C. General Guidelines............................................................................................................................................................... 2 D. Detailed Guidelines.............................................................................................................................................................. 1 1. User Interface:...................................................................................................................................................................... 1 2. Accuracy Considerations...................................................................................................................................................... 1 3. Fluency Considerations........................................................................................................................................................ 5 4. Stylistic Social Media Considerations................................................................................................................................... 5 E. References........................................................................................................................................................................... 6 Introduction These project guidelines contain comprehensive information about translation guidelines. Linguists are requested to read all the topics in detail before working on this project. Project Overview In this project, linguists will be required to: - Correct a machine translation of a short speech transcription. The source is comprised of audio & text (the speech and transcription, respectively); only the text requires translation. Translation Guidelines A. Terminology a. Source (src) language is the language of the provided text. b. Target (tgt) language is the language to be translated into. c. Transliterate is to represent or spell the sound of the source text word(s) in the target language alphabet. d. Post is the translation unit consisting of source audio file, source transcribed text, and suggested machine translation. B. Assumptions a. Translator needs to be at a minimum native in the assigned non-English languages (exceptions may be made for rarer languages), aware of the cultures of both languages, and understands the nuances of social media language. b. We assume that translators are able to translate technical content using appropriate terminology where warranted. c. We assume that translators possess the capability to generate accurate translations by listening to the source audio, referring to the accompanying source transcription, and referring to the suggested machine translation to generate accurate translations. C. General Guidelines a. Provide fluent translations for the provided post (referring to the transcription and machine translation suggestion and optionally listening to the audio) to the target language without deviating too much from the source structure; only allow necessary changes. b. Do not expand or replace information, do not add any explanatory/parenthetical information, definitions, etc. c. Do not ignore any meaningful information that was present in the source audio and transcription. d. Do not translate [x] or [c] tokens seen in the transcript and just copy them over in the corresponding location for their translated counterpart. e. In the case that the audio and corresponding transcription have text in another language but does not meet the reject criteria. Try to translate if possible, if not possible/ language is not understood just copy through to the translation. f. In case of multiple possible translations available, please pick the one that makes the most sense (e.g., for gender concordance, cultural fit in the target language, sociopragmatic linguistic register [level of formality], etc.). g. Please provide translations that are faithful to the source in terms of pragmatics such as (if present) level of hedging/modality, sentiment and its intensity, negation, speech effects (disfluencies), etc. h. We suggest using Urban Dictionary as a reference where appropriate/needed. i. If the given machine translation is perfectly translated, ensure that the exact provided text is submitted. Do NOT leave a blank text box, a substitution character (such as -), or a message stating that no correction is needed. j. Reject a job for the following reasons: i. Source has more than one intelligible speaker: More than 50% of the audio contains two or more speakers speaking at the same time. This includes musical lyrics being sung over someone speaking. ii. Source contains more than 10% in a different language: Audio is in a language other than that of the queue you are transcribing or the user’s accent is such that you are unable to understand the utterance (this includes “child speak”). iii. Source has more than 75% unintelligible speech. iv. Source has no speech to translate: Ensure the entire audio clip has no speech before rejecting. D. Detailed Guidelines 1. User Interface: a. Instructions (blue section), Translation Input box (section below instructions), Reference Audio, corrected transcript and suggested machine translation (to the right)  2. Accuracy Considerations Objective: render the most accurate equivalent translation possible preserving meaning and cultural nuance. Examples are provided in tables illustrating confusion points with incorrect translations in red, dispreferred in orange. a. Localization/Internationalization i. Numbers: If the number is spelled out, translate it to the equivalent form in the target language, if it exists. Source Wrong Target Correct Target (EN) That was back in 1949. (NE) त्यो 1949 मा फिर्ता थियो (NE) त्यो १९४९ मा फिर्ता थियो (EN) Three beers please. (HI) ३ फियर कृपया। (HI) र्तीन फियर कृपया। Table 1A: Numbers transliterated into target script ii. Unit of measure: Do not convert any unit measurements (Don't convert °C into degrees Fahrenheit, nor miles into kilometers, nor pounds into kilograms, nor inches into centimeters, nor currencies). Source Wrong Target Correct Target (EN) I need 2lbs. of rice. (IT) Ho bisogno di 1kg. di (IT) Ho bisogno di 2 libbre di riso. riso. (AR) ‫ كيلو تفاح‬٢ ‫أريد شراء‬. (EN) I would like to buy 4lbs. (EN) I would like to buy 2kg. of apples. of apples. Table 1B: Unit of measure iii. Time: should be translated to the equivalent form in the target language and not necessarily literally. Source Wrong Target Correct Target (EN) It is 9:30 p.m. (IT) Sono le 9:30 p.m. (IT). Sono le 21:30. (AR) ‫ م‬٩٫٣٠ ‫الساعة‬. (EN) It is 9:30 in the evening. (EN) It is 9:30 p.m. Table 2: Numeral Time iv. Dates: For both numeral dates and spelt-out dates, please localize them according to your local culture. Source Wrong Target Correct Target (EN) She arrives 5/6/20. (IT) Arriva il 6/5/20. (IT) Arriva il 5/6/20. )AR(.١٩٥١ /٦/٥ ‫ولدت هدى في‬ (EN) Huda was born on (EN) Huda was born on 05/06/1951. 06/05/1951. Table 3: Numeral dates Source Correct Target (EN) She arrives on the 6th of May, 2020. (IT) Arriva il 6 maggio, 2020. Table 4: Spelled out dates v. Diacritics: if present in target language, please make sure to specify diacritics accordingly to ensure the meaning is preserved: Source Wrong Target Correct Target (EN) Going to make tea (ES) Voy hacer te (meaning (ES) Voy hacer té (té with the literally: Going to make a T - diacritic mark - tilde - means T is spelled as te in spanish) tea) (EN) What a campaign (ES) Que campana (meaning (ES) Que campaña (the ñ literally: what a bell) distinguishes the meaning to be campaign) Table 5: Importance of diacritics b. Code Switching/Mixing i. Replace instances of code switching/mixing with [c]. For multiple words in sequence that are code switching/mixing, write [c] only once. Also be cognizant of code switching vs. borrowed words. Code switching involves switching between different languages in speech, whereas borrowed words are specific terms from another language that have become part of the new language’s vocabulary (see example 6b). Source Correct Target (EN) I want to go to my abuela’s house. (ES) Quiero ir a la casa de mi [c]. Table 6a: Code switching/mixing Source Wrong Target Correct Target (EN) His forte is (ES) Su forte es la (ES) Su punto fuerte es la communication. comunicación. comunicación. Table 6b: Borrowed words: originally of foreign origin, but are now part of a language’s vernacular  c. Idiomatic Expressions and Proverbs i. Idiomatic expressions and proverbs should not be translated literally if possible. Translators should either use an equivalent idiom, if one exists, or they should use an idiom of similar meaning. If no equivalent or similar expression exists in the target, translators should paraphrase the source idiom so that the meaning is transferred to the target language. d. Song lyrics and Poetry i. If a post contains spoken song lyrics and/or spoken poetry not attributed to background music they should be translated. If there is an official translation, please use it. Otherwise, translate it as usual rendering an equivalent translation. It is acceptable to translate literally even if the rhyming is lost. e. Sarcasm and Irony i. Sarcasm and irony should be translated into the same social register. Source Wrong Target Correct Target (EN) No shit Sherlock! (IT) Niente merda Sherlock! (IT) Non mi dire Sherlock! Table 7 Sarcasm and irony f. Gender Bias in Translation: (shortcoming: acknowledging that we are assuming a binary gender setting) i. Pronoun translation: When translating from a language that is underspecified for pronouns in the absence of context, for example a 3rd person pronoun “it” to be translated is ambiguous and could be interpreted as either masculine or feminine, translators should do a 50% of the time translate into feminine and 50% translate into masculine. Source Correct Target Correct Target (EN) They want to be taller. (IT) Vorrei essere più alta. (IT) Vorrei essere più alto. (EN) They bought it for him. (AR).‫اشتروه له‬ (AR).‫اشتروها له‬ Table 8: Pronouns: Example of either feminine or masculine, both valid but don’t render both, please pick one. g. Professional Titles i. Some languages have gender reflected in professional titles similar to actor/actress in English however if the title is ambiguous as to whether the referent is male or female, use judgment if the rest of the post indicates a specific gender or approach similar to (i) above by translating into feminine 50% of the time and masculine 50% of the time. Source Correct Target Correct Target (EN) Dr. James is competent (IT) Il dott. James è (IT) La dott.ssa James è and personable. competente e simpatico. competente ed è simpatica.  (AR) ‫الدكتور جيمس كفوء و لطيف‬. (AR) ‫الدكتورة جيمس كفوء و لطيفة‬. Table 9: Professional titles: Example of either feminine or masculine, both valid but don’t render both, please pick one. h. Profanity i. Source containing profanities should always be translated into equivalent rather than the literal translation. If the profanity is extremely offensive but does not meet the criteria specified in the rejection reasons, please translate to the closet possible equivalent Source Wrong Target Correct Target (EN) Damn it! (IT) Maledizione questo! (IT) Maledizione! Table 10: Profanity i. Understandable and common computer language or vernacular phrases i. If there are equivalents in the target language that should be the first preference (e.g. LOL, OMG) otherwise translate into the correct meaning of the phrase. Source Wrong Target Correct Target (EN) ROTFL (IT) ROTFL (IT) Rotolarsi a terra dalle risate Table 11: Abbreviations used on social media Source Wrong Target Correct Target (EN) Ima chill out (IT) Ima freddo fuori (IT) Mi rilasso Table 12: Vernacular j. Proper Nouns i. Proper nouns include people's names, toponyms (place names like New York City, or China), movie and TV series titles, venue and event names, product and brand names, game and team names, band and song names, and musical genres. Please leave proper nouns in English if there are no official/common translations in your language. If common/official translations or transliterations of proper nouns exist, please use them. Source Wrong Target Correct Target (EN) We are going to (IT) Andiamo a Florence (IT) Andiamo a Firenze Florence this summer! quest’estate. quest’estate. Table 13: Proper nouns that have a common translation in target  Example in which the proper noun remains unchanged in the target language because 1) it is easily recognizable in target language; and, 2) there is no other frequently used or known translation. Source Wrong Target Correct Target (EN) I love Lady Gaga! (IT) Adoro Signora Gaga! (IT) Adoro Lady Gaga! Table 14: Proper nouns that should remain unchanged ii. In the rare case that transliterating that proper noun to the target script is so uncommon that it would almost be incomprehensible to the local speaker, then the translator can use their best judgement to keep in the original script. k. Abbreviations i. Typically they are translated into the full name or transliterated depending on the language and context. Follow guideline, for Proper nouns (j). Source Wrong Target Correct Target (EN) USA/US and the WHO (IT) USA e WHO (IT) USA e l'OMS Table 15: Abbreviations of named entities 3. Fluency Considerations Objective: Produce the most accurate translation NOT to copy over incoherent language a. Sound/Speech Effects i. Sound/speech effects such as stutters should not be translated. Source Correct Target (EN) Th-the sky is b-blue (ES) El cielo es azul Table 16: Translation of sound/speech effects 4. Stylistic Social Media Considerations Objective: Maintain fidelity to the original source language a. Links, Emojis, Mentions/Usernames i. Always preserve links and make sure they appear the exact same way in the target as in the source!a b. Elongations and Contractions i. Elongations and contractions in source should be reflected in target language where possible and appropriate. Source Wrong Target Correct Target (EN) I luuuv you! (IT) Ti amo! (IT) Ti amooo! Table 17: Translation of terms that are elongated E. References a. https://www.atanet.org/publications/Getting_it_right.pdf