Computer-Aided Interpreting (CAI): The Most Unnoticed Innovation of 2017
Interpreters have historically used a variety to tools to prepare themselves for interpreting assignments. However, their technology choices have been limited to an interpreter console, interpreting delivery platform, or glossary. That’s about to change with the emergence of full-fledged computer-aided interpreting (CAI).
What is CAI?
CAI refers to computer software that supports professional human linguists in the preparation and delivery of accurate interpreting assignments. Such technology helps them be more efficient and systematic, which is crucial to manage the wealth of information to prepare for events and to recall relevant terminology while interpreting. The nascent CAI category will provide three modules that help linguists:
- Prepare for interpreting assignments. Among other things, this capability includes a database of support material, the ability to synthesize such documents automatically, terminology extraction features with a database to store candidate terms, and term memorization widgets.
- Assist during the delivery of interpreting assignments. This refers both to manual term lookups and automated ones based on voice recognition.
- Support processing post-assignment information. This speeds up tasks such as linguistic asset maintenance, quality assessment, and reporting functions.
InterpretBank’s Termbase Interface
Why Is CAI Only Appearing Now?
Interpreters have used ad hoc tools to prepare for events and sessions for years. However, until now, most have re-purposed products, in particular terminology tools that weren’t designed for interactive use during interpreting assignments. As a result, many interpreters found them unreliable or too complex to use in the midst of a job. Instead, they continued their pen-and-paper-based information management.
What’s new and different is software that goes beyond manual lookups during interpreting sessions. InterpretBank recently released a version of its product in which a voice recognition system transcribes what the speaker says. It then performs automatic terminology lookup, displaying terms and their translations on screen. It uses a termbase that the interpreter prepared ahead of the session. The tool also shows numbers and proper names to reduce the note-taking strain. This marks the beginning of dynamic technology assistance during sessions, the second module in the CAI armory.
InterpretBank’s Automated Term Lookup
The product also helps interpreters prepare for their sessions, the first function for CAI software. Its core functionality supports: 1) collecting corpora and extracting linguistic information; 2) creating and managing glossaries; and 3) providing tools for memorizing terminology.
InterpretBank did not start as a CAI tool. Its founder, Claudio Fantinuoli, had the idea to develop software for interpreters to manage terms and knowledge so they can manually look up this information during interpreting sessions, as well as prepare systematically for conferences. While interpreters already take manual notes on problem triggers – such as names or numbers – manual lookups tax cognitive capacities that are needed to focus on the actual interpretation.
Fantinuoli’s true innovation was to automate the lookup to reduce the cognitive load. He did that by enabling his tool to search for terms in transcriptions generated by voice recognition software such as Dragon Naturally Speaking. However, a big caveat at this point is that interpreters still need to take their own notes as a backup in case the voice recognition doesn’t pick up the name or number.
Where is CAI Going?
InterpretBank is the result of a PhD project at the University of Mainz/Germersheim. The size of the interpreter pool is too limited to justify large investments and many practitioners do not push for technology to assist them out of fear of being replaced by it. As a result, the sector isn’t likely to see major developments that result in commercially available solutions. However, some interpreting delivery platforms (IDPs) already use or are developing automatic term lookup in their systems, and more may do so but for their own advantage.
The current technology has barely scratched the surface of what can be done. CSA Research expects to see an eventual explosion of tools to help interpreters before, during, and after interpreting assignments, whether for simultaneous interpreting, on-site interpreting, or remote interpreting. Further developments in the coming years will include a broad range of features. There will be integrations with interpreting management systems (IMSes) for the coordination of assignments and with remote interpreting platforms for use of the tools within the interpreting delivery platforms. Also, the addition of capabilities such as automatic content enrichment (ACE) will enrich the preparation process by providing on-the-fly background information to prepare terms for the assignment.
Successful providers of CAI solutions will need to be relentless in their usability testing to ensure interpreters become more effective in their preparation work rather than being disoriented by a wealth of distracting information. Interface design will require usability testing in the field to perfect how linguists prepare and consume information while interpreting. In addition, developers will need to build partnerships with educational institutions to boost the training of new interpreters. Tools will be more likely to take off if new practitioners are introduced to these capabilities in school and come to expect them in their work environment.
About the Author