[unable to retrieve full-text content]
YouTube automatic subtitle translation not working issue addressed PiunikaWebWith the adoption of expert-in-the-loop translation steadily growing and becoming the dominant production method, translators have seen their productivity increasing.
Depending on their skills, language pair, content, and machine translation (MT) engine, translators typically achieve significant productivity gains, as representatives of language service providers (LSPs) and post-editors told Slator during interviews carried out for the Machine Translation Expert-in-the-Loop Report.
In his keynote address at SlatorCon Remote June 2022, Translated CEO, Marco Trombetti, said translators work now at faster speeds than ever before. But how fast can they actually work at postediting machine translation output?
According to a 2022 blog post by Karina Martinez Ferber, Product Manager at Acolad, a single translator without any MT aid can deliver, on average, 2,000 to 3,000 words daily — a measure historically seen as a solid median value since the mid-1990s.
By adding machine translation postediting (MTPE) into the mix, the word count doubles, reaching up to 7,000 words in a single day, said Zsófia Lelner in a 2022 post by translation technology provider memoQ.
A view also shared by Across. According to a 2021 blog post by Flurina Schwendimann, former Marketing Content Manager at Across and freelance translator, “post-editors process some 7,000 words daily.”
Talking to Slator, Enrico Antonio Mion, translator, post-editor, and postediting trainer said, his throughput is 1,000 words per hour for light postediting, and 700 words per hour for full postediting. Adding up the numbers, the result confirms the estimates above, as Enrico would turn around 8,000 words per day and 5,600 for full postediting.
In terms of performance patterns, he noted that his productivity usually peaks after using the same MT engine for three months, “as that’s when you get used to the MT engine, you can anticipate the error, and you actually work faster than translation [without MT].”
Productivity is mainly affected by the language pair, the content, the MT engine, the requirements regarding the target text, and the post-editors’ skill set, as mentioned in the MT Expert-in-the-Loop Report.
For example, in a 2019 productivity test of domain-adapted neural machine translation (NMT) carried out by Textshuttle’s CTO Samuel Läubli et al., the professional translators of Migros Bank’s internal translation team delivered 700 words per hour on average when postediting financial texts.
Similarly, around 700 words per hour were delivered by the post-editors who participated in a 2018 study by Antonio Toral and Martijn Wieling from the University of Groningen and Andy Way from Dublin City University when postediting a novel that had been translated by an MT system tailored to novels. (The authors observed a large variability in production speed: from the lowest value of 402 words per hour to the highest of 1,140).
Five professional translators working for the Barcelona-based LSP Incyta delivered on average 1.500 words per hour when postediting news articles that had been translated by their own commercial MT software, Lucy Software, according to a 2020 research paper by Sergi Alvarez Vidal and Antoni Oliver from Universitat Oberta de Catalunya and Toni Badia from Universitat Pompeu Fabra.
Drawing data from Acolad, it would be fair to suggest that “Limit Up” in performance is an equation of different variables. That is, the easier the content, the better the MT system, the lower the target-text requirements, and the higher the post-editor’s skill set, the higher the possibility of reaching the Limit-Up words per hour.
In fact, in a 2018 research paper by Olena Blagodarna, Associate Professor at the V.N. Karazin Kharkiv National University, the answers of more experienced post-editors were skewed toward the segments of 900–1,000 words per hour and ≥1,000 words per hour.
This was further reinforced in a 2019 Slator article, where the top limit increased to a maximum of >5,000 words per hour, with 50% of post-editors scoring between 1,000–3,000 words per hour.
Nov 25, 2022; Milwaukee, Wisconsin, USA; Milwaukee Bucks guard Grayson Allen (12) drives against Cleveland Cavaliers forward Isaac Okoro (35) in the first half at Fiserv Forum. Mandatory Credit: Michael McLoone-USA TODAY Sports
Packers bye week mailbag: Difficult decisions ahead
Using film to break down NBA games is more prevalent on social media than ever before. That means you must keep up with all of the acronyms such as DHO, SLOB, and ATO that get thrown around all willy-nilly. Luckily for you, the Milwaukee Bucks Film Room Dictionary will help you become a smarter fan.
Every Sunday, I’ll break down a different term and give video examples about what it looks like. Last week we dove into Sideline Out of Bounds (SLOB) plays and how the Bucks use them to build upon each other. This week, I define dribble handoffs (DHO).
The goal is to help you become a smarter basketball fan and understand the game at a higher level.
The format of each video will be as follows:
Dribble Handoffs are a normal part of the Bucks’ offense this season and are important to know. Head coach Mike Budenholzer doesn’t like to run a ton of set offense, but he does have certain rules in place. One of them appears to be that a DHO action is initiated when the ball-handler dribbles toward a player in the corner. This opens up many options for the offense, gets the defense moving, and creates advantageous opportunities for the Bucks.
These DHOs are an integral part of the Bucks’ offense. It helps keep the flow going when they aren’t in set plays and ensures good ball movement. Milwaukee likes to give everyone on the court an opportunity to score the rock, and this creates opportunities to do just that. Keep your eye out for these moving forward.
A new term will be released in the dictionary every Sunday. Please feel free to comment on the YouTube video about future terms you want to see defined.
Setting up your company in a new country offers a range of exciting possibilities to expand your business. However, your business will not be able to capitalize on new opportunities without a localized marketing campaign and without a strategy to tackle language barriers.
Customers prefer to read reviews in their native language, buy products from websites that speak their language and read information that they comprehend.
On top of that, your business needs to function in the new environment, which necessitates dealing with hiring, accounting and other processes in a new language. Ultimately, your business needs a way of handling translation on a holistic, centralized level.
While artificial intelligence for language translation is a powerful tool to create efficiencies within your business, it’s absolutely crucial for building trust with consumers. The kind of language a brand uses defines the customer-brand relationship and also establishes a long-term connection.
Successful businesses are using an integrated approach that uses existing tools, such as machine translation, in combination with human editors. This approach can improve organizations’ marketing strategies and processes. It also can help businesses understand and communicate more effectively with new audiences, grow customer relationships and increase brand credibility.
When considering expanding into a new market, companies should understand the language of their new customer base well enough to generate new leads. One tool that is quick and efficient as part of a marketing strategy is the chatbot.
Chatbots act as the frontlines of communication without requiring the full localization team to be online at all times. When new customers browse your site and need help, or if they’ve been looking at the same product page for more than a minute, chatbots can reach out to the customer and answer any questions or concerns instantly.
Conversational commerce – which means using online customer service and chatbots to offer personalized shopping experiences and drive sales – exploded in the e-commerce industry because of its ease of use.
You can take it a step further by implementing an AI translation layer. Through AI, chatbots will be able to answer questions and upsell products to potential customers in any language.
Another way of incorporating AI translation technology is by collecting customer questions to the chatbots, along with the appropriate responses, to create comprehensive FAQs for the website, which can then be made available to customers in different languages.
As your business expands to new markets, you need to build marketing campaigns that appeal to your target audience’s preferred tone and style. The use of an AI translation solution provides a nuanced approach to language and can help you avoid a big marketing faux pas.
You also can use AI translation to understand how a native target audience reacts to an ad campaign. Other possibilities include conducting surveys and polls to glean data about the new market, and translating these insights to improve the messaging for your social media platforms, content marketing and ad campaigns.
Large corporations like Mastercard, Deloitte and Credit Suisse already use AI marketing tools to automate content creation, which allows them to boost productivity by creating content in less time. The Associated Press uses AI to write corporate earnings reports, freeing up reporters to do more in-depth reporting. You can apply an AI translation layer to the marketing tools you already use to quickly and efficiently write content that appeals to the new market, in a tone and style that is sensitive to the preferences of native speakers.
With market expansion, it’s difficult to communicate with a range of potential new stakeholders within your company, whether they are new marketing employees, PR people, or sales representatives native to the market. Use of machine translation tools can lead to a holistic approach for multilingual businesses, streamlining internal processes to ensure alignment with stakeholders, particularly when it comes to your business messaging and strategy.
There are many examples of marketing use cases where AI translation can be employed. It can be used to translate and maintain product catalogs, product documentation, price lists and other sales collateral for sales representatives.
Also, if your company’s communication team is drafting a media announcement, they can have it translated quickly for outreach to all targeted publications in their respective languages – while ensuring the messaging is sensitive to the target group’s preferences.
A translation tool also can facilitate the creation of detailed briefs for product managers in their own languages to help them gather, manage and prioritize market and customer requirements. The list of ways to aid your marketing efforts by integrating an AI-based translation layer is a long one, and the tool can be employed wherever there is content to be translated.
Note that AI translation doesn’t replace the marketing function in a business; rather, it advances the efficiency of the marketing team. When businesses set up shop in a new country, the usual approach is to employ large and unwieldy localization teams that only scale based on how many people are on the team. Incorporating a small team of human editors to refine an AI-based translation will help you conserve your resources.
Using an AI-based translation layer, marketing teams can easily target and attract new customers, because marketers are able to focus on crafting new strategies without being tied down by the translation aspect. As AI improves, customer expectations increase accordingly. You can ensure your marketing campaigns are ahead of the curve by leveraging AI technology for fast and efficient translation.
Vasco Pedro is the co-founder and CEO of Unbabel, the AI-powered Language Operations platform that helps businesses deliver a multilingual customer experience at scale. Vasco is a seasoned expert in AI and language technologies. After a degree in language and knowledge learning, Vasco went on to develop his master’s degree and doctorate around language technologies. He then worked at both Siemens and Google, where he helped develop technologies to further understand data computation and language.
Subscribe to SmartBrief’s free newsletter to get the latest marketing news and insights direct to your inbox.
OpenAI recently unveiled the AI conversation interface ChatGPT. ChatGPT is a large language model using GPT-3.5 technology and has been trained on a massive amount of data — including, apparently, manual finetuning and training. ChatGPT generates human-like text in response to a given prompt. The launch set the AI world on fire and OpenAI netted a million new users within days.
After Slator tested the waters by asking ChatGPT questions about the language industry, we took it a few steps forward by producing a video with ChatGPT, Synthesia, and Descript.
First, we asked ChatGPT to answer questions as if they were a translation manager on a podcast. Here are the unedited results.
Slator: Hey, welcome to this podcast. You’re my guest today.
ChatGPT: Thank you for having me on your podcast. It’s always a pleasure to be a guest and to have the opportunity to share my thoughts and ideas with others. Do you have any specific topics or questions that you would like to discuss during our time together? I am here to help in any way I can.
Slator: Yes, please pretend to be a Translation Manager at a large ecommerce firm. In two sentences, what’s your job day-to-day?
ChatGPT: As a Translation Manager at a large ecommerce firm, my job is to oversee the translation of our company’s website, marketing materials, and other content into different languages. This involves coordinating with in-house translators and external translation agencies, ensuring that translations are accurate and consistent, and managing our translation budget. On a day-to-day basis, I might be working on translation projects, reviewing and editing translations, or communicating with other members of the team and our external partners to ensure that everything runs smoothly.
Slator: Ok, that’s interesting. What’s the single biggest challenge for you in your job?
ChatGPT: One of the biggest challenges in my job is ensuring that the translations we produce are accurate and of high quality. This can be difficult for a number of reasons. For example, different languages have different grammatical structures and conventions, which can make it challenging to accurately convey the meaning of the original text. Additionally, there can be cultural differences between languages that can affect how a translation is perceived by speakers of that language. Ensuring that our translations are accurate and of high quality requires careful attention to detail, a deep understanding of the languages we are working with, and a willingness to seek out additional resources and expertise when needed.
Slator: Do you have team members that can help you?
ChatGPT: As a Translation Manager, I likely have a team of in-house translators and external translation agencies that I work with to produce translations for our company. Working with a team can be beneficial because it allows us to divide the workload and leverage the different strengths and expertise of each team member. For example, I might have team members who are fluent in multiple languages and have experience translating different types of content, such as marketing materials or technical documents. By working together, we can ensure that our translations are accurate and of high quality, while also meeting deadlines and staying within our budget.
Slator: Is there a language combination you think is particularly difficult to find qualified linguists for?
ChatGPT: There are many language combinations that can be difficult to find qualified linguists for, depending on the specific languages involved and the type of content that needs to be translated.
Slator: How do you use machine translation?
ChatGPT: As a Translation Manager, I might use machine translation in a number of different ways. For example, I might use machine translation to quickly produce a rough translation of a text that can then be reviewed and edited by a human translator. I might also use machine translation to help identify potential translation errors or inconsistencies, or to quickly translate a large amount of text that would be time-consuming or difficult for a human translator to handle. Overall, the use of machine translation can help to streamline the translation process and make it more efficient, while still ensuring that the final translation is accurate and of high quality.
Slator: To wrap up here, how do you think your job will change over the next five years?
ChatGPT: I expect that the use of machine translation and other technology-assisted translation tools will continue to grow and become more sophisticated, which could change the way that we approach translation projects. Overall, I expect that my job will continue to evolve and change over time, and I will need to adapt and stay up-to-date in order to remain effective in my role.
Slator: Thanks a lot for joining today. Goodbye!
ChatGPT: You’re welcome. It was a pleasure to be a guest on your podcast. Goodbye!
With the script ready we then used Synthesia, an AI video generation platform, to turn the text into a video. All you had to do was copy the text generated by ChatGPT and paste it into a new video in the Synthesia app. No downloading or uploading is needed. Once an AI avatar was selected, the video was processed and then downloaded.
For an extra step, we moved onto audio and video editing platform, Descript, to customize the video with a background and captions. The final product was exported and uploaded to YouTube.
While this final result may not fool everyone, this combination of AI tools takes ChatGPT one step further to being able to impersonate a human.
Most advice is worthy of thorough examination. Often advice about one discipline — with just a bit of translation — has meaning in other disciplines. This translation process allows for more perspective and greater insight.
As a case in point: The more I study improvisation, the more parallels I see to the business world. And while I’ve written about this practice in the past, I continue to discover more connections. So what follows is even more improv advice in translation.
Improv advice: Don’t take any shortcuts on energy or polish. You may have done the show a thousand times, but some people see it for the first time. Business advice: It’s showtime every time.
Improv advice: Play with improvisors of various experience levels. New folks will remind you to make clear offers. Veterans will keep you in your emotions. Business advice: Strive for connectivity at every level within your organization. New hires will look to you for help navigating company culture and expectations. Tenured leaders will help you to understand the business from various perspectives.
Improv advice: Cast cohesion takes time. No matter how funny you or your troupe-mates are, you must be patient while the culture comes together. Business advice: Team cohesion takes time. No matter how brilliant everyone is, learning how to work seamlessly and succeeding collectively takes time and patience.
Improv advice: Conflict in a scene can take many effective forms. One versus one. One versus many. Everyone on stage against the world. Business advice: Healthy conflict can serve as a catalyst to strengthen teams and relationships.
Improv advice: Keep an improv journal. Write down ideas for shows. Write down good feedback you’ve been given. Write down good feedback you’ve overheard. Business advice: Keep a journal and spend time in reflection. Write down the lessons you’ve learned along the way.
Improv advice: See as much improv as you can. Constantly remind yourself that your style is one of many and that you exist as part of a huge, diverse art form. Business advice: There are a lot of ways to run a business. There are a lot of ways to lead. Your style can be unapologetically all you.
Improv advice: Side-coaching from your director or troupe-mates doesn’t mean you’re bad at improv. It means that you’re good enough for them to think you can get even better. Business advice: Hard to hear feedback from a peer or supervisor can be challenging. It can also be exactly what you needed to know in order to become a better version of yourself.
Improv advice: Find out how to perform to a cold crowd without compromising your format. Find out how to perform to a rowdy crowd without compromise too. Business advice: Adjusting your style for a given situation is not the same thing as compromising who you are fundamentally. Flexing is good; selling out is not good.
Improv advice: Different formats will make you improvise differently and so will different troupe-mates, different directors and even different performing stages. One of the goals as an improvisor is to understand how different configurations of people, places and things impact you as an artist. Business advice: Different is different. And what’s different needs to be taken into consideration.
Improv advice: There is no amount of funny you can be that justifies being a bully, being unprofessional or being arrogant. In improvisation, trust is foundational to execution. If you make yourself a burden to work with through your behavior, you simply cannot do good work. Business advice: Trust and moral character matter. Full stop.
Improv advice: The world is big enough for everyone’s comedy. Business advice: The world is big enough for everyone’s contribution.
Here’s the funny thing I’ve noticed: The more parallels I see to other areas of my life, the more I want to study improv.
And I can’t wait to keep learning.
Ancora Imparo… (Still, I am learning)
Stacey Mason is the founder of The Improv Lab, a professional development business in Bentonville. More information is available by calling 479-877-0131. The opinions expressed are those of the author.
Here is a dilemma: How do our students who are new to our country learn when they don’t understand the language?
It’s something our English as a New Language (ENL) teachers deal with every day. They keep looking to improve the educational experience of our ENL students in any way they can, making the transition as seamless as possible.
Kate Shadrin, a special education teacher at CVMS, read an article recently about a teacher who applied for a grant to buy Timekettle WT2 Plus Translator Earbuds, a device that translates verbally and in writing what a teacher is saying in class and, in turn, what a student replies.
Ms. Shadrin brought the article to Maureen Curtin, the secondary ENL leader. Members of the ENL department knew of such devices but thought they were cost prohibitive. Now, Ms. Curtin realized, maybe there was a way to get funding for them. That started the ball rolling. Ms. Curtin talked to Principal Lee Hankinson who was onboard with the new devices.
“Principal Hankinson has always been so supportive of our ENL students,” said Ms. Curtin. “She immediately utilized building funds to make this opportunity for our newcomers possible.”
How the earbuds work is the classroom teacher wears one earbud and the student wears the other. The technology is also attached to the student’s Chromebook. As the teacher speaks, the student hears the lesson in his or her language of choice. On the Chromebook, the text comes through in both English and the student’s specified language. There are 40 languages to choose from with 95 percent translation accuracy.
“The earbuds are an amazing tool. Our content-area teachers can use them to translate material into other languages,” said Ms. Curtin. “The translation takes about two to three seconds. It is heard in their native language, and written in their native language as well as English.”
Instruction is always provided in a student’s first language, she added, with translated textbooks and materials. This technology is huge because it’s immediate and in the classroom.
For many newcomer students, English is their second – or sometimes their third – language, according to Ms. Curtin. This device not only helps them learn and understand this new language, but also helps the social/emotional piece by decreasing anxiety. They are understanding and can pick up with learning immediately. The students can also participate in the class discussion.
“Our goal is to have students use their first language while they learn English. This is called translanguaging,” said Ms. Curtin. “The earbuds are a great educational support for this. They allow the students to participate and be an active member of the classroom. The technology will also help teachers determine if there are any learning deficiencies.”
The devices were used first at Circleville Middle School in a pilot program with R.J. Gabriel in science and Abby Williams in math. Normally, the instruction is all in English. For the student that knows little or no English, the time spent in class is not always productive and sometimes frustrating. These devices help students learn from the start. Hope Schaumburg, an instructional technology coach for the secondary schools, was instrumental in the rollout, helping with trouble shooting along the way, along with Ceia Acvedo, the CVMS instructional technology assistant.
More devices were then delivered to Crispell Middle School and Pine Bush High School.
The advantages of these devices go beyond the classroom. Ms. Curtin said teachers can take a screenshot of the text and email it to families so that they can help their children at home, creating a home/school connection from the start for ENL families. The devices can even be used during parent/teacher conferences.
“This is so exciting,” said Ms. Curtin. “It’s yet another tool for us to help teach our students. The goal is to see a lot of growth and for our students to achieve success.”
She certainly has great expectations for this technology.
“This is going to be a game changer.”
