Machine Translation Master Course Newsletter – Issue 5
We are adding new features to the Machine Translation Master Class, 2022 edition!
Since the Machine Translation Master Class was launched in October 2020, students from all over the world have taken it. As the instructor, I thoroughly enjoyed working with them and was extremely grateful for their valuable feedback and inputs. This is a good beginning! In 2022, we are adding some new features.
MT as Part of NLP AI
Naturally, a machine translation master class is about machine translation. However, MT does not live in a vacuum. It works closely with other supporting NLP tasks embedded in a localization process. Thus this class takes a holistic approach to MT, examining it in the bigger picture of the whole localization ecosystem. From a technical perspective, we can look even deeper. NMT is based on artificial neural networks, driven by machine learning and deep learning.
A neural network language model aims to predict what word comes next by assigning probability to a piece of a text sequence. It has a very high level of autonomy and is less constrained by humans. In front of an artificial neural network, all problems are reduced to an input-output one. It’s all about data: if you feed the model with parallel corpus, it captures the alignment/ equivalence patterns in translation; if it is a corpus of paired questions and answers, it learns to answer questions; and if it contains long texts and their short summaries, it aims to accomplish a text summarization task.
By looking under the hood of NMT, we actually unlock the immense power of neural networks from within and have access to other NLP tasks. Yet this class aims to help you develop a habit of thinking to incorporate NLP tools and methodologies in your work, rather than developing computer programs yourself.
The value added by humans in an automated translation process
When MT becomes more powerful, a question often asked is: “What values can we humans add?” To answer this question, we need to revisit the nature of translation with MT as a new dimension. Is achieving equivalence between the source and target texts the only reason why we localize content? If so, a machine translator is working hard to deliver this relationship. Soon not everyone will be suitable for the task of working with machines for this purpose. Many more need to go beyond this stage and expand the scope of their tasks, for example, linguistic analysis of MT or MTPE results for quality control, MT customization, MT quality estimation, style and cultural considerations, communication effects and terminology preferences.
The ELIS 2021 European Language Industry Survey reports that most successful candidates for an in-house language function had a Master in Language, not a specific Master in Translation or Interpreting. Today many MT companies start to build cumulative NMT, which combines multiple natural language tasks to produce better translation results. These are all signs for us to consider what areas to prioritize in order to add our unique human values in an automated translation process. With a NLP AI mindset, we are able to find our own way that fits us best.
Rethink operational practices
In an MT implementation process, buyers and suppliers have different considerations. However, the boundaries become more blurry due to the impact of AI. For example, specialized suppliers, typically those technology companies, start to expand their business scope from simply offering tools to a complete solution that includes both manpower and technologies. Traditional language service providers are developing their own MT and NLP systems to be more competitive. This class will incorporate this dynamic feature when discussing use cases and the practicalities of MT solutions. We will summarize some Do’s and Don’ts when implementing MT tools and cover topics such as data security and privacy.
- MT does not live in a vacuum. It is part of the family of NLP AI
- Develop a NLP AI mindset to find your way to add unique human values
- MT AI is shaking up the market, in particular, on the supplier side
Looking forward to seeing you at the Machine Translation Master Class, 2022 edition!
Disclaimer: Copyright © 2021 The Localization Institute. All rights reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published, and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this section are included on all such copies and derivative works. However, this document itself may not be modified in any way, including by removing the copyright notice or references to The Localization Institute, without the permission of the copyright owners. This document and the information contained herein is provided on an “AS IS” basis and THE LOCALIZATION INSTITUTE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY OWNERSHIP RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.