Machine Translation Master Class Newsletter – Issue 5 – January 2022
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 humbly collected their feedbacks along the way: (1) going a bit more into practicalities of existing MT solutions; (2) balancing the content to meet the needs of class participants with various backgrounds; and (3) encouraging conversation with the other course participants. Their valuable suggestions are the reason why we are making some changes in 2022.
Feedback #1: Rethink Operational Practices
To address feedback #1, the second session will emphasize MT implementation based on knowledge of the inner workings of multiple MT paradigms. Then session 4 will revisit this topic from the angle of risk management. It focuses on humans’ role to leverage various resources for translation quality control, as well as the coordination between MT and its supporting techniques.
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 systems to be more competitive. We will incorporate this dynamic feature when discussing use cases and the practicalities of MT solutions in our class.
Feedback #2: Stakeholder Collaboration
Class participants have various backgrounds, being primarily a translator, a project manager, a localization architect, and a decision maker, to name just a few. To balance the teaching content for them is a challenge, yet when we consider the collaboration among stakeholders, this task is manageable. After all, no matter what role one plays in a localization process, they are all humans.
Naturally, early progress in MT started in the technical world, who focused on sketching out the mathematical framework to tackle the problem of translation in their own way. The famous quote by Frederick Jelinek in 1998, “Every time I fire a linguist, the performance of the speech recognizer goes up” showed that this approach worked two decades ago. However, in the new era of AI, simply relying on quantity alone to get the low hanging fruits no longer applies to harder, more subjective problems. In particular, in the MT deployment stage, what really makes a difference is human users, who give meaning and purpose to it by customizing and designing it. Collaboration among people with different roles is crucial for MT deployment. This class will focus on the common ground of MT implementation shared by stakeholders, including the technological principles behind MT, methods for quality control as well as Do’s and Don’ts in specific use scenarios.
Feedback #3: Prioritize Human Learning
Feedback # 3 is absolutely valuable. Learning is, really, to find out what we already know and to explore what we can know. To better help participants learn, we will implement a new class format, that is, each session will have a 10 minute break between two 45-minute periods, followed by 20 minutes of Q&A. In addition, we will use an e-learning platform to facilitate communication among class participants as well as that between participants and the instructor.
I firmly believe that the fundamental solution for practitioners seeking to leverage AI is that humans have better learning capabilities than machines. In other words, we will feel confident in using it only when we know we can beat computers in learning. Even though many aspects of machine learning in translation will be covered, human learning will and should be prioritized in our class.
Looking forward to seeing you at the Machine Translation Master Class, 2022 edition – you can sign up here!
Some might wonder how the Machine Translation Master Class is different from another class I am offering in March 2022: Ready-Set-Go THINK NLP. If so, please check out the following table to compare their features:
|Machine Translation Master Class||Think NLP RSG class|
|In common||Both classes are on technology, but prioritize humans. They approach translation and localization from the perspective of a linguist, a project manager and other domain expert, as opposed to a technical expert such as a computer programmer, both of whom are main drivers leveraging AI to magnify and scale human knowledge of translation.|
|Differences||● It focuses on machine translation as well as other tasks associated with MT implementation.
● It includes both the conventional CAT tool environment and innovations MT can bring about.
● It emphasizes humans’ role in MT quality evaluation, estimation and overall quality control.
|● It does not prioritize any particular application, but tells the story of neural network language models.
● It includes multiple NLP tasks that have been or can be applied in a localization process.
● It depicts the role of localizers in data collection, annotation, management and application.
|Key Takeaways||● Knowledge of multiple MT paradigms
● A perception of what innovations MT could bring to a CAT tool environment
● An understanding of humans’ role in MT quality evaluation, estimation and overall quality control
● An understanding of how to coordinate between MT and its supporting techniques
|● Fundamental NLP AI knowledge and methodologies
● An understanding of how to get insights from data and metadata as well as a localizer’s role in this process
● A NLP mindset to help you think proactively to leverage NLP AI in your business
● A perception of how to effectively motivate and manage your man power in an AI driven localization process
Dr. Peng Wang’s Previous Newsletters
About the Author
Dr. Peng Wang has taught, researched and practiced translation and localization on three continents. She is the convener for EDUinLOC, the chair of the automation/AI track for LocWorld conferences and a part-time professor at the University of Ottawa. Previously, she was the CAT Tools Coordinator at the University of Maryland.
Dr. Wang has rich research experience on NLP AI, having worked as a linguistic researcher at the Corpus Research Lab at Northern Arizona University, a domain expert for data mining projects at the University of Maryland and an honorary research fellow on automatic discourse analysis tools at the University of Liverpool. Her current research and practice focus on human learning vs. machine learning, machine translation, terminology and multilingual corpus analysis.
Dr. Wang is an expert in approaching technology in the context of culture and humanities. She embraces linguistic and cultural diversity in her classrooms, with students aged from 18 to over 70, in over 10 language combinations, coming from UAE, China, Italy, Spain, Germany, Morocco, Colombia, Mexico, and Haiti, to name just a few.
Connect with Dr. Peng Wang:
Connect with Dr. Peng Wang on LinkedIn: https://www.linkedin.com/in/pengjanewang/.
Contact Us - Machine Translation
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.