1 week ago
July 12, 2019
Reader Polls: MT and Subject Matter Expertise, Future CAT, Outperforming LSPs
Each week, we poll the more than 10,500 subscribers to Slator’s weekly email newsletter, asking them to weigh in on what is top of the mind in the industry in the context of recent news.
At the end of June, we wanted to know if subject matter expertise is becoming even more important (at the expense of source language proficiency) as neural machine translation increasingly does a lot of what some call the grunt work of translating words.
Many language service providers (LSPs) are experimenting with workflows where subject matter experts do light post-editing of machine translated content, mostly looking out for obvious errors in terminology and apparent flaws from a simple reading of the target language without having to refer back to the source.
Our readers are fairly evenly divided over whether better MT does indeed drive the tradeoff of source language proficiency in favor of subject matter expertise, with 43% disagreeing, 39% agreeing, and 17% saying it depends.
Someone who is firmly in the yes camp is Trados creator, Coreon CEO, and newly appointed Board Member of Polish LSP Summa Linguae, Jochen Hummel. In an interview on his recent appointment, Hummel told Slator that NMT’s “human parity changes everything, because ‘doing words’ is done by a machine. However, for several use cases, revision by subject matter experts is required.”
But how will linguists work with all this increasingly high-quality machine output? In a report presented at the International Annual Meeting on Language Arrangements, Documentation and Publications (IAMLADP) forum held in Brussels from May 27–29, 2019, the European Union said that “language technologies and AI ideally take over the mundane part of language professionals’ work.”
The EU, one of the world’s largest buyers and users of translation, said that most of its translation services have now integrated its eTranslation service into their pre-processing arrangements, and provide NMT output to their linguists. That said, the EU still sees room for improvement in terms of usability and emphasized that “the future smarter CAT environment is yet to be developed.” Over 90% of the Slator readers we polled agree.
Investors, meanwhile, continue to believe in the future of language services. So much so that the small group of publicly listed language service providers handily outperformed the broader stock market since the beginning of the year.
Leading the charge are Australia’s Appen and Tokyo-listed Rozetta. But RWS and Keywords Studios, too, put in a solid first half of 2019, while Honyaku Center and SDL struggled, and ZOO Digital nosedived after one of its main customers changed supply chain operations.
We asked our readers to do some crystal ball gazing and over half expressed confidence that the language industry’s good run is set to last for the next 12 months.
Join SlatorCon San Francisco on September 12, 2019, to hear Mark Brayan, CEO of Appen, discuss How to Make Language-based AI Solutions Work in the Real World.