| { | |
| "paper_id": "1981", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T07:48:58.113901Z" | |
| }, | |
| "title": "", | |
| "authors": [ | |
| { | |
| "first": "Barbara", | |
| "middle": [], | |
| "last": "Kostrewski", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "The City University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Peter", | |
| "middle": [ | |
| "Taylor" | |
| ], | |
| "last": "Aslib", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "The City University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "The following points were made in the discussion of papers by Green and Lavorel (read by McCluskey) on the pilot use of French-English and English-French Systran in the Commission of the European Communities, Luxembourg.", | |
| "pdf_parse": { | |
| "paper_id": "1981", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "The following points were made in the discussion of papers by Green and Lavorel (read by McCluskey) on the pilot use of French-English and English-French Systran in the Commission of the European Communities, Luxembourg.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "McCluskey, in delivering Lavorel's paper, had noted that what he himself found most irritating was the unnecessary transposition of \"not-found words\", e.g. \"American Bible Society\", occurring in a French source text, appeared in the English target text as \"American Society Bible\". He emphasised that the aim should be accuracy (100%), intelligibility and style (the criteria identified by Green), and that postediting should be learned very thoroughly. While pressures in industry and commerce might be such that mistakes had to be tolerated, the Commission could not work to this principle.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "1.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The use of MT is being tested in Luxembourg only, and not in the Commission's translation divisions in Brussels. Reliable figures are not yet available, but MT may perhaps now account for 3% of the Commission's translation load in Luxembourg, and 1% of the total load including Brussels.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "2.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Ruffino contrasted the relative luxury of the Commission's very selective Systran experiment with the situation in industry and commerce, where urgent largescale translation was normal: faced with the need to translate a manual before a product can go on the market, a translator is delighted to have an aid which may lift his throughput overnight to 150 pages a day. The Commission, on the other hand, was faced with a unique diversity of translation tasks, often with restricted subject expertise on the part of the translator, and no indication of the translation's ultimate purpose. This complexity made careful experimentation essential.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "3.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Wheeler (CEC) stated that errors of substance would get through only if the translator regarded what comes off the machine as automatically correct. The posteditor should always work with the original text; if errors occur, this is because of bad translation practice.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "4.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Whereas Commission posteditors at present work on hard copy, their changes are finally entered on word processors.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "5.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Commission posteditors' throughput varied widely with the type of text, up to 30 pages a day for a general text if the translator knows what the user wants. The quality of final product which the posteditor seeks to achieve may be quite different from what the user requires. Much MT output may be nowhere near acceptable to human translators, but nevertheless acceptable to the user.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "6.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Concentrated emphasis on detail in postediting was said to be a luxury which commercial concerns could not afford; the machine translation would have to be good", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "7.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "enough or need only minor changes. It may therefore be necessary to establish criteria for acceptable translation, particularly when a large volume of material had to be translated. The need for technical accuracy was also stressed, being more important than correct grammar.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "SESSION 3/SUMMARY OF DISCUSSION (1 )", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Since the time taken for a translation might be critical the decision on whether to use MT or not could be left to the translator; the improvements based on posteditors' feedback were not allowed to delay delivery of the final translation, but were incorporated in the system shortly afterwards. A fundamental principle was to avoid anything that interfered with the normal translation routine. Green noted that, because of the uniquely wide range of subjects to be translated in the Commission, the task of incorporating feedback was necessarily large, complex and ongoing, whereas in business most of the feedback might be performed in a single operation after installation of the MT system.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "8.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The distinction was made between the postediting of raw machine translation by a translator/posteditor, and the revision of translation (whether human translation or postedited MT) by a reviser. In the Commission translation division in Luxembourg, all MT is both postedited and revised. Ideally, editing and revision should not be performed by the same person. Indeed, it was suggested that postedited work should not need revision in any case, since posteditors are senior staff.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "9.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "It was emphasised in discussion that machine translation can speed up translation by several orders of magnitude, depending on the complexity and extent of postediting.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "10.", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": {}, | |
| "ref_entries": {} | |
| } | |
| } |