eCoLoRe Translation Tasks

1
eCoLoReTranslation Tasks

• Karl-Heinz Freigang, Uwe ReinkeFachrichtung
  4.6"Angewandte Sprachwissenschaft sowie
  Übersetzen und Dolmetschen", Universität des
  Saarlandes, Saarbrücken

2
Overview

• Design guidelines for creation of materials
  (cooking)
• Learning Support Materials
• Providing terminology
• Files to exercise TM tool (Déjà vu)
• Files to exercise localisation tool (Passolo)
• Timetable (for Q2 and Q3 2003)

3
Design guidelines (cooking of
materials)General remarks

• Different learning goals require different
  material and thus different recipes for
  material creation
• introductory modules for TM LION newcomers
• modules for advanced learners (testing/evaluating
  TM systems)
• LSP translation courses integrating TM tools
• project-oriented courses for advanced learners
  (regarding both tools and LSP translation)
• different types of working environments
  (freelancer vs. service provider (in-house staff,
  freelancers))

4
Design guidelinesGeneral remarks

• Cooking of texts to exercise evaluate fuzzy
  matching should start from translators (human)
  understanding of similarity
• rough distinction of the following cases
• SL SLTM (exact match)
• same content repetitions
• different content ambiguities
• SL ? SLTM (fuzzy match but also no match)
• same content paraphrases
• without shift of information from text to
  extra-linguistic context (and vice versa)
• with shift of information (explicitation,
  implicitation)
• modified or related content
• more detailed diversification of cases required
  (? will become language-specific at a certain
  level of linguistic concreteness)

5
Design guidelinesIntroductory modules for TM
newcomers

• focus on introduction to major functionalities of
  TM systems
• material either unmodified real-life texts or
  cooked texts
• texts should not be too long but must contain
  some amount of exact and fuzzy matches(maybe
  parts taken from longer texts)

6
Design guidelinesModules for advanced learners

• should contain material that makes learners aware
  of the shortcomings of fuzzy match and alignment
  algorithms
• texts should be comparatively short
• focus on evaluation

7
Design guidelinesModules for advanced learners

• Material for evaluating fuzzy match algorithms
• cooking of text material distinction of
  typical matching scenarios
• material might encourage learners to create
  individual test suites
• TM populated with data from preceeding project
  (update scenario) focus on exercising problems
  of pre-translations (i.e. TM should also contain
  ambiguous material)

8
Design guidelinesModules for advanced learners

• Material for evaluating alignment algorithms
• cooking of SL-TL text pairs to be aligned
  include alignment scenarios that usually cause
  problems
• segmentation problems caused by
• shortcomings of alignment algorithm
• typographical/DTP problems in SL and TL texts
• nm-alignments (n?m n?0m?0)

9
Design guidelinesLSP-translation courses with TM

• should ideally use unmodified real-life material
• texts should not be too long so that learners can
  work on several texts (text types) per
  term/semester(maybe meaningful parts taken from
  longer texts)
• focus on translation task, LSP terminology etc.
• requires either
• empty TM (from-scratch scenario)
• or
• TM populated with data from preceeding project
  (update scenario)
• cooking of text material only if no predecessor
  texts available

10
Design guidelinesProject-oriented courses

• should ideally be based on unmodified real-life
  material (ideally complete texts)
• focus rather on workflow and project-management
  issues than on individual functions and
  evaluation of TM
• requires either
• TM or unaligned texts of preceeding project
  (update scenario)
• or
• empty TM (from-scratch scenario in this case
  material must contain sufficient amount of
  internal repetitions)
• cooking of text material only if no predecessor
  texts available

11
Design guidelinesSuggestion for structure of the
cookbook

• 0. Introduction (purpose of the cookbook)
• 1. Describe different target groups and learning
  goals
• 2. Specification of relevant text types
• 3. Text cooking recipes for different learning
  goals

12
Learning Support Materials (LSM)

• LSM needed for
• Admin functions (materials analysis, basic
  project management functions)
• QA functions (e.g. proofreading TM, maintaining
  TMs, terminology check)
• Alignment (problems of sentence-based alignment
  like sentence segmentation and complex SL-TL
  correspondences, problems caused by DTP errors)
• Terminology extraction

13
Providing terminology

• Creation/extraction of terminology
• a) providing terminology as part of the eCoLoRe
  training material (? materials design issue)
• b) collecting/extracting terminology as a
  learning goal (? learning support material issue)

14
Providing terminologyeCoLoRe training material

• obstacle no commonly supported exchange format
• provide term lists in the form of tab-delimited
  files and/or Excel files
• should contain synonymous terms, contexts,
  source, infos on client, project etc.
• compiled manually or with the help of term
  extraction tools
• material should cover the following issues
• terminological issues (homonymy in LSP, synonymy,
  client-specific terminology)
• retrieval issues
• make learners aware of term recognition problems
• include situations where TM doesnt yield matches
  but TDB does

15
Providing terminologyInstruct learners how to
collect/extract project-relevant terminology

• Should this be part of eCoLoRe?
• We think No - at least not yet, because this
  would require a complete introduction to
  (translation-oriented) terminology management

16
Files to exercise TM tool (Déjà vu)

• Original SL texts for translation (and
  monolingual term extraction)
• TM files in TMX format to be imported into DV
• Original SL and TL texts for alignment (and
  (mono- or bilingual) term extraction) and/or
  bilingual TM

17
Files to exercise localisation tool (Passolo)

• SL files containing interface resources (EXEs,
  DLLs maybe resource files)
• Glossaries
• ready-made material like MS glossaries
• project-specific glossaries
• SL and TL files of previous version(s) (EXEs,
  DLLs resource files) to practice leveraging
  process
• Other file formats?
• Further files to practice complete localisation
  project (e.g. help files)?

18
Timetable