An innovative medical learner’s dictionary translated by means of speech recognition

By November 17, 2016,
Page 497-500
Author Arnoud van den Eerenbeemt
Title An innovative medical learner’s dictionary translated by means of speech recognition
Abstract I will be discussing a medical dictionary based on the Keyword in context (KWiC) concept and speech recognition as a valuable tool for both speeding up data input tasks and preventing overstrained hands and arms while using the keyboard. The latter is vital for me as my daily full-time work consists of compiling medical dictionaries for students and professionals (creating and updating complex and dynamic data) and creating medical spellcheckers.
Non-Anglophone medical students and health care professionals around the globe need an active command of professional English for their career. Yet the lexical tools available for acquiring these skills are few and insufficient: American and British explanatory dictionaries expect readers to be native speakers, while bilingual medical dictionaries are basically glossaries and provide unlabelled translations. The only medical learner’s dictionary in the world to date is the excellent Fachwortschatz Medizin, a comprehensive dictionary compiled by Michael and Ingrid Friedbichler. It helps non-native speakers to acquire language skills step by step, is structured using modular medical concepts and combines various lexical features:
· monolingual dictionary: 100,000 medical terms grouped into 1400 sections with key headwords defined in simple English; contextualized with collocations and sample sentences demonstrating correct use, extracted from a 20-million-word corpus of medically authoritative texts;
· semi-bilingual dictionary: support in the user’s native language (German, Dutch) in the form of 42,000 translated keywords;
· thesaurus: synonyms, antonyms and related terms;
· domain-specific glossary: readers from all medical fields can focus on content relevant to their specialization;
· PC edition: full-context search, customizable display (pronunciation, definition, translations, collocations), cross-references etc.
After acquiring the Dutch rights I realised out that farming out the translation work would require me to extensively monitor translators. I decided to translate the 42,000 medical terms myself instead, using speech recognition and a large monitor to display my database, a web browser, a word processor and two medical dictionaries. I developed voice-driven macros for automating 600,000 Google searches, creating 2000 records, searching dictionaries and my 52,000-record medical database etc. This allowed me to translate up to 400 terms per working day and complete this immense task within reasonable time.
Session The Dictionary-Making Process
author = {Arnoud van den Eerenbeemt},
title = {An innovative medical learner’s dictionary translated by means of speech recognition},
pages = {497-500},
booktitle = {Proceedings of the 14th EURALEX International Congress},
year = {2010},
month = {jul},
date = {6-10},
address = {Leeuwarden/Ljouwert, The Netherlands},
editor = {Anne Dykstra and Tanneke Schoonheim},
publisher = {Fryske Akademy},
isbn = {978-90-6273-850-3},