VIA Programs

Graphics localization

Altogether, we localized 20 graphics for VIA Programs into 3 locales–zh-CN, zh-TW and ja-JA. These graphics include infographics in program introduction pages and marketing graphics.

Our client created their graphics using Adobe Illustrator, and we received the original .ai files. However, we didn’t receive packaged files, but individual files instead, as shown in the picture on the right. This resulted in many missing links and fonts.
While we were waiting for our client to get back to us with all the missing links, we came up with an alternative solution. Because that the graphics are all created for web publishing purposes, these graphics don’t need to be in high-quality. So the alternative solution was that we use Photoshop to recreate the text-boxes. We did eventually get the missing links from our clients, so I localized the graphics using Illustrator. Also, since there were too many graphics to deal with, and the only way to localize them with Illustrator was through manually copying and pasting all the texts and create picture-lists, I thought of recreating all the graphics in Indesign since many CAT tools support .indd files. But this resulted in many misplaced texts and images, and the CAT tool we were using, SmartCAT, didn’t support INDD files (but will be in the future as indicated on their website). So creating picture-list was the only solution with regard to graphics localization.

Website Localization

All together we localized 12 web pages into 4 locales–zh-CN, zh-TW, ja-JA and th-TH.

Our client built their website using WordPress. This left us with two choices in terms of the tools that we choose to make their website multilingual: WPML and q-TranslateX, two WordPress plugins that enables website localization. A major difference is that WPML creates separate posts for each language, while q-TranslateX gathers contents for different languages in one post. We communicated with our client about the choice of tools, and their goal was to make the web pages multilingual but at the same time, the viewers would be able to quickly go back and forth between English and their native language for references. Q-TranslateX became a better choice, and taking into account that it’s free and both our client and ourselves have experience with the plugin, we were both happy about the choice.

The client trusted us enough to let us modify the original website. This was both a blessing and a challenge–although it’s efficient and straightforward to work on the original website, it was also terrifying because using q-translateX meant that I had to manually copy everything from the website. While I was doing that, I was afraid I might accidentally delete texts from the original website. My solution towards this challenge was that I kept screenshots of the original websites and documented every change I made on the website. Also, I mock-built every page for every language to make sure the page would be successfully localized into the dedicated language.

The other challenge I had localizing their website was the website templates. The templates can not be localized using q-TranslateX, as shown in the picture below. I downloaded a plugin called “Export Plugins and Templates” in WordPress and exported the templates. Using global search in TextWrangler, I was able to locate the template that was used to create the pages that need localization. The template was created using PHP, so the next step would be for me to internationalize the PHP file and import the PHP file back to the website. Before implementing the practice I consulted with our client, but they were concerned about the internationalized templates tampering with the original design, and they thought the idea of leaving some words in English was acceptable. So further discussion is needed on this practice and we are planning on having another meeting with our client hopefully before we graduate.

We are the localization team of our company XYZSoft. We have developed a software in two languages, English and Spanish. Now, our software is going to the Arabic-speaking market, but our product team keeps writing UI strings with placeholders, concatenation, spelling errors, and incomplete sentences. We have told them many times that they should write the codes that meet the internationalization (i18n) standard but they wouldn’t listen. We, the localization team, understand well that in order for a smooth localization process to happen, especially when the target language is an extremely tricky one, we need to evangelize our product team to modify the codes in our product.

In order to do this, we put together a presentation to our product team. We started out with an overview of our case and explained why there haven’t been many critical feedbacks so far. Then, we listed out some of the problematic coding issues that would lead to disastrous results when the product is being localized into Arabic. We also provided some examples of i18n coding conventions for our engineers in the product team to refer to. After that, we illustrated how by improving our i18n efforts we will save resources and cut budgets in the long run. An updated workflow was then shown to them, with i18n integrated into the early stage of the product development cycle, as shown in the graphic to the right. Finally, we showed our product team how quality will be improved and maintained by implementing a new workflow and involving Language Service Providers.

Some reflections on our presentation:

What went well in our presentation:

1. It was a great teamwork I’ve missed for a long time. We knew our part by heart, and we knew each others’ parts as well. We modified the connections between each member’s contents several times to make sure our presentation flows well and that no one part of our presentation seems isolated.
2. In order to explain the case well, we added some details to this somewhat abstract case. We listed specific numbers to illustrate the problem about i18n which seem more straightforward and at the same time more persuasive.
3. We provided the detailed workflow that targets specifically at i18n not being in the early stage of the product development cycle.
4. The Sharks (i.e. our product team) were satisfied with the answers we provided towards their questions (although for the most part our answer was “we will discuss this later in our presentation”).

What did not go so well:

1. In my opinion, we should have involved the Sharks more, since our goal was to persuade our product team about our new practice.
2. Just before our presentation one of our team members raised a question: So have our products been localized before? Yes, it has been localized into Spanish–this is a question that should have been brought up in the very beginning.

What you would have done differently:

1. To involve the Sharks more, we could have addressed some issues directly to the Sharks, or raise questions for them.
2. Should have invested some more time in the design of our slides. They are not as important as the contents, but it would certainly add value to your presentation. The way that Team Case 4 had designed their slides was impressive and made the contents more appealing.

What I have learned:

1. Make sure everybody understands the end goals of our presentation in the very beginning before going into specific details about how this goal could be achieved.
2. Teamwork can be pleasant if there’s a clear goal to achieve, and your teammates are all as passionate about the cause as you.

Our team, Loong Localization, has taken up two projects for the course “Localization Practicum” with professor Max Troyer offered by Middlebury Institute of International Studies at Monterey. For the first project, we are localizing Mojito, an open source automation platform for continuous localization developed by Box. For the second project, we are helping VIA Programs, a non-profit organization that offers innovative experiential learning programs in Asia and the United States for people around the globe.

Better Tools, Better Life 

The tools that have made our life much easier in this case are:

• Email: for communication with the clients
• Asana: for Project Management
• WeChat: for communication
  
• Box: for internal documentation and collaboration with Box team
• SmartCAT: for translation and  vendor management
• Google drive: for document sharing, internal documentation

Project I: Mojito

We are helping Box to localize the Mojito website and web app into 8 different languages, which are, Chinese Simplified, Chinese Traditional, Japanese, Korean, Spanish (Latin America), Italian, German and Portuguese.

The Web App

All the files have already been extracted as .xliff files by the Box localization team from the web app.

Lessons learned from the way Box markup their .xliff  file:

• ‘Id’ and ‘resname’ are important attributes that point to the location of the string in the original resource. When vendors have questions concerning a translation unit, it should be reported to the client with the corresponding id and resname.
  

  A glimpse of the source .xliff file.

• There’s slight difference between ‘id’ and ‘resname’. “The id attribute is used to link a <trans-unit> or an inline element to its original location in the source file from which the XLIFF document was produced. The id attribute values are determined by the tool that created the extracted document, they may or may not be the same as the values of the resname attribute. The resname attribute holds the original identifier of the text item extracted in the <trans-unit> element.” (https://wiki.oasis-open.org/xliff/FAQ)
• When a unit is translated and/or validated, the ‘state’ of the unit should be changed. Most of the CAT tools change it automatically, including SmartCAT.
• ‘Note’ is a comment for the translator to better understand the context of the unit.

Progress

So far, we have already received the translated files for four languages for the web app. Because that the Mojito website is mostly about instructions on using the web app, we have decided to extract terms from the translated web app to create a glossary, which will be one of the project deliverables. One minor issue to pay attention to is that some of the terms may appear as part of an attribute or variable name, as shown in the screenshot below: 

Project II: VIA Programs

For this project, we are localizing seven sections of VIA’s website, five program videos, as well as a number of brochures into four languages, namely, Chinese Simplified, Chinese Traditional, Japanese, as well as Thai.

The Website

VIA Programs built their website in WordPress. After our discussion with our client and having made sure that they have their website backed up, we have decided to use q-Translate plug-in to translate their website. The main reason is that our client wants the visitors to be able to refer to the original English website when necessary, and q-Translate allows us to add language pickers on the English website, for free. WPML offers the same solution, but with a price. We also expressed concern that activating q-Translate and making modifies on the original website might affect the layout of it or even result in some of the sections being destroyed. But our client gave us the go ahead on the original website, stating that they already have the original website backed up and can reset it if necessary. However, they want us to document any changes we make. Therefore, we are doing documentation for this process in a word document, as suggested by our client. 

The Infographics

A website filled with data and plain text bores the audience. VIA Programs team is obvious well aware of this and has featured about twenty infographics in their Programs Introduction sections.  We have received the original files of the infographics,  which contain the original Adobe Illustrator (AI) files, exported PNG files, and images they used to create the AI files. However, the files we received were not the original package generated by Illustrator, which should include the AI file itself, a Link folder with all the images used, and a Fonts folder with utilized fonts. Taking into consideration that the client has minimal experience with localization and to avoid misunderstanding of what items are actually missing, I compiled an Excel document with the missing file names in one column and the reference images in the other, as shown in the screenshot.

Progress

For the web texts, we have thus far extracted all the texts into word documents. The corresponding translation tasks have also been assigned to our translators on SmartCAT. With regard to the infographics, the texts have been externalized and converted into picture lists. However, if our client is unable to locate the original images and fonts, we may try to localize the infographics in Photoshop by creating text masks on the background layers.

For the final project for the course “Software & Game Localization” that I took at Middlebury Institute of International Studies at Monterey, my team (Bosi Lin, Jenny Lowe and myself) localized a web-based javascript game “Infectors” into Traditional Chinese.

Infector: the Chosen Game

Finding the right game to localize was a challenge. We browsed through this website of open-source javascript games,  played many of them and decided on “Infectors”.
The game has a nice and clean UI with a “Gameboy” touch to it, has a complete story line, and is neither too hard nor too easy to play. It has a picture (the title) on which we need to do some desktop publishing work, and it contains the right amount of strings. Therefore we downloaded the source codes and tested the game. Except for the fact that the game was built on a particular server and we were only able to run it smoothly in Firefox, no bugs and errors showed up when we tested it.

Overview of Localization Workflow

Internationalization

Bosi was responsible for the externalization of our strings. He also managed to add a language picker to our game.

Localization

As I was preparing the strings for translation, I encountered a few issues. The first one was the comments generated by our engineer Bosi. While the comments make the strings organized, it caused trouble when I tried to use the “Advanced Find and Replace” function to hide the strings from being recognized by CAT Tools. As you can see from the screenshot below, the best I could do was to find the forward slashes and the first word after. The best solution would be to add two slashes after each comment, so that the comments could easily be found and replaced automatically.
The second issue was that when I tried to email the string files and picture list to our translator Jenny, the email got rejected by the server several times. The reason for this, we soon discovered, was that the file has the word “infector” in its name and thus was detected by the server as virus. Anyways, lessons learned–there’s a reason why professor Troyer kept emphasizing that we should always avoid transferring project assets through email. It may seem easy and convenient at first, but don’t get deceived by it’s appearance!

Testing

The biggest headache we had when testing the game was that the Chinese characters wouldn’t show up. This is a serious issue because the characters were pretty much the only reason why the game needed localization. So, after some research and consulting we figured out that the original fonts in the game were actually bitmap fonts, as you can see from the screenshot below.

Bitmap font is essentially a kind of font where each character is stored as an array of pixels. It’s not easily scalable as opposed to vector fonts. And bitmap font easily works with English because English is a language with only 26 alphabets. Chinese, on the other hand, uses a completely different system to form its characters.
And now we know that the only solution to this issue would be to find a bitmap font for Chinese characters–except we couldn’t. So we decided to create a bitmap font only for the texts that show up in the game.

Finalization

Since we changed the font, we had to make sure that every .js file that has font information written into it has a Chinese version with the new bitmap font written into them. The files we had to create are shown in the screenshot below.

And we ran the game again to make sure that everything runs smoothly without any errors.

A sneakpeak of our localized game.

Introduction

This portfolio serves as an overview of the course “Translation Management Systems” I took at Middlebury Institute of International Studies at Monterey. The main purpose of this course is to help us understand comprehensively, with lots of hands-on practices, different types of translation management systems, including WorldServer and XTRF. We also looked at how the translation management systems are integrated into the workflows of localization processes. Based on all these knowledge we acquired in class, I worked with four team members to localize a website into three languages, with WorldServer being our main management system. The related documents of this project will also be presented in this portfolio.

TMS Project Files: Localizing the The Gathering Place official website

Project Proposal

Deliverables

Presentation on Lessons Learned

How Can Translation Management Systems Help
Translation Crowdsourcing?

TMS Selection: Things to consider before investing in any TMSs

Introduction

This is a project showcase for a course “Multilingual Desktop Publishing” I took when I was pursuing a master’s degree at Middlebury Institute of International Studies at Monterey. The course is mainly about the hands-on experiences on tools most used in the desktop publishing process, such as InDesign, Photoshop, Illustrator, After Effects, Sigil, etc. This project showcase is for us to demonstrate one of the skills I acquired from this course. For the source file, I chose the map of Universal Studios Florida, which localized into Simplified Chinese.

Workflow

In order to get a general idea of the time scope needed to complete the project, I used Wunderlist to create a simple checklist of tasks with deadlines assigned to each of them.

Finding the right resources was not as easy as it may seem. I finally settled on the map of Universal Studios Florida in PDF, which Acrobat showed me was originally created using InDesign CC 2015, so I used PDF2ID to convert the document into INDD format. Unfortunately, the conversion messed up the formatting, a lot. As you can see from the screenshot, the icons are flowing up and down, and the alphabets in the black boxes are misplaced. It would take hours to get all the small icons into their right places.

Discouraged, I decided to try Illustrator instead. It turned out that for this project Illustrator did a better job of converting the PDF, as you can see from the screenshot below. 

To clean up the document, I first had to find the missing fonts for the source text. Then I brought up the guidelines and aligned the text boxes.

After making sure that there’s nothing culturally inappropriate, I went on to prepare the document for translation in Trados. In addition to all the texts on the screen, the icons with English texts need to be replaced. Furthermore, the numbers in inches in some small icons need to be converted into centimeters.

After translating all the texts, I did some reformatting such as adjusting the text boxes to fit the Chinese characters.

Challenges and Best Practices

1. For each line, the space between the icons and the text box before them has to be the same (or at least similar). This took some nudging using arrows on the keyboard. The best practice would be to put one or two extra characters in the text box and nudge it to contact with the left box. The extra charactes could be deleted afterwards. 
2. For the small icons for which I needed to replace the numbers, because that many of them have different numbers in it–some are two-digit, some are three–it was hard to make them all look exactly the same. But considering that they appear small on the map, it doesn’t affect much the overall consistency of the map layout.
3. Some of the names appear on the map don’t have official Chinese equivalent. To solve this problem, I looked for the map of Universal Studios Japan which had a version in Traditional Chinese, as a reference.
4. After importing the map into InDisign, it warned me of the missing fonts. Taking a screenshot of it helped me a lot when I was finding fonts in Chinese text, for this way I know how many fonts were used on the original map, in which part, and for what purposes.
5. I left the copyright part untranslated (the map of Universal Studios Japan remained this part unaffected), so I had to make sure it looks the same as the original document. In the PDF document,the texts were well wrapped but it’s not the case anymore in Illustrator. Therefore, I recreated a text box and paste the contents into it. This way I could easily wrap it up and make adjust alignments with a single click.
6. For some texts that appear several times, like “Quick Service” or “Full Service”, can be localized using “Find and Replace” to save time. 

Introduction

This portfolio serves as an overview of the course “Advanced Computer-Assisted Translation” I took at Middlebury Institute of International Studies at Monterey. The main purpose of this course is to help us understand comprehensively different types of productivity software that language professionals use today. However, this is not at all a course packed with theories–we had a lot of hands-on practices using a variety of software such as SDL Trados, Microsoft Translator Hub, Okapi Olifant, Okapi Checkmate, and so on.

In this portfolio, I have included three projects completed throughout the course.

Microsoft Translator Hub:
Machine Translation Project

DeepBLEU-Statement of Work

DeepBLEU-Statement of Work Updated

DeepBLEU-Presentation on Lessons Learned

Custom Filter Exercise in SDL Trados

For this exercise, I customized the filter settings in SDL Trados for XML files using a sample file, a screenshot of which is attached below.

This is an XML file with a large number of elements that need not be translated. To prepare such a file for translation, setting up a customized filter inside Trados can save a lot of time and effort for everyone.  Here are three screenshots I took for the filter I set up for this specific file.

For this exercise, understanding the hierarchical relations between “Always translatable” and “Not translatable” is the key. To illustrate, because there are so many elements within this file and the majority of them doesn’t need to be translated, I set the RSS element, which is the master element of the file, to “Not translatable”. But before that, I also remembered to make the elements that have contents need to be translated “Always translatable”. The system automatically prioritizes the “Always translatable” rule over “Not translatable”. This way, all the elements will not be seen by translators in Trados except for the one made “Always Translatable”.

Here is a screenshot of the pseudo-translation results after applying the filter to the file.

Future of Translation Technology

Future of Translation Technology

Introduction

This online portfolio is an overview of and a reflection on the course “Social Localization and Translation Crowdsourcing” I took in Middlebury Institute of International Studies at Monterey, taught by professor Adam Wooten. Translation and localization crowdsourcing is a trend. Nowadays not only large technology companies like Google and Facebook are incorporating crowdsourcing efforts into their translation and localization products, small companies and start-ups have also been looking and experimenting for best practices in the same field. In this portfolio, I have included some suggestions and evaluations of social localization and translation crowdsourcing efforts based on what I have observed in and out of the course. Furthermore, a link for a project I have completed with a team is attached at the end of this portfolio. 

I. Maintaining Quality in Volunteer Translation

The surge of social network and the continuously growing need of localized products has been the main reasons why crowdsourced translation has become a trend.  In this community-based collaboration process (most of the time voluntary),  how could a company/organization maintain the quality of their translation? Here are a few tips that may come in handy when managing volunteer translations.   

Clearly State Your Purpose

First of all, before launching the translation process, the senior manager must have a clear purpose in mind. That is to say, the manager knows their organization’s target market and knows what kind of message needs to be conveyed. For the localization of ourrescue.org, for example, the main purpose is to make the organization known to more people, therefore attracting more to donate and support. A clear purpose of a localization process is crucial to deciding what kind of translators are needed, whether they should pass certain tests, etc.

Assign a Supervisor

A supervisor is not always needed in crowdsourcing projects. Nevertheless, it is strongly recommended that an employee is chosen to keep an eye on the whole process. Sometimes the supervisor can even be a professional linguistic. For ourrescue.org, considering the nature of the organization, it is best that they use one of their members to manage the translators.

Provide Guidelines and/or Reference Tools

A simple guideline should be provided before translation kicks off. It can be a checklist of several dos and don ts and/or a style guide. A glossary or even a translation memory can also be useful.

Encourage Communications and Implement Voting System

In translation phase, timely communications between the manager and translators are vital in ensuring quality translation, in that if feedbacks are provided and reviewed in time adjustments or new guidelines can be offered should there occur any critical errors. Voting is another mechanism to ensure translation quality and has been widely used by many organizations with their crowdsourcing translations. Volunteer translators who have received a considerate amount of votes can be considered as a “super voter” who can later perform a final review for the translations.

Final Review   

After the translations are done, the results should be further reviewed. This process can be carried out by translation machine and/or the “super voters”. For ourrescue.org, for example, in order to save time and effort, three or four translations conducted by the same translator can be deemed as representative of his/her work. If the representing pieces are of good quality, then the reviewers should jump to the next translator and let the machine do the rest of quality check.

II. Marketing to Recruit Volunteers for NetEase

This is a proposal for NetEase’s open course website to better market to recruit volunteers, who will be translating the subtitles of their widely popular open courses online.

Target Your Market

Before any marketing campaigns take place, a company should make clear what their target market is. What NetEase needs are translators that can translate subtitles mainly from English to Simplified Chinese. In that case, the target market would be people who have enough knowledge of English and Chinese to perform the subtitle translation. The target market would best be people who are the audience of the open courses themselves so that they would be more dedicated to the cause. However, some of them are interested in sharing information and knowledge. By doing so fulfills their sense of accomplishment. Therefore, they can be listed as one of the target markets.

Set the Goals

The goals of the marketing campaigns are to maintain the translator that are currently working on translations as well as to attract more potential volunteers to devote themselves to the cause. Moreover, through the campaigns, NetEase should expect to gain reputation and seize public attention.

Decide the Marketing Mix

Products

The products that NetEase is offering would mainly be the sense of accomplishment, in that the videos to be translated are mainly for educational purpose. Some volunteers with would also see this as a way of practicing their translation skills.

Price

In the case of translating videos, the prices for volunteers to consider are opportunity costs, i.e. what they could have achieved with the same amount of time used in translating subtitles. The overheads is another expense for volunteers, but it wouldn’t be a major concern for either the volunteers or the company. NetEase could consider providing the volunteers with online privileges, certifications or gifts as compensations. Signed books or T-shirts are also highly

Place

NetEase could consider putting the email address of the manager in charge of recruiting translators as a watermark on their videos. They could then consider cooperating with some of the major video streaming websites such as youku.com and iqiyi.com to stream their open courses. By doing this, whoever has seen the videos can easily get in direct contact with the recruiting team.

Promotion

There are several ways for NetEase to promote themselves. They can go to volunteer subtitle translation groups that already exist to recruit. They can also put up information on social media. In addition, they could even go to campus to promote their projects, for young people tend to be enthusiastic about volunteer works, and it is much more likely that they are the audience for these open courses themselves.

Plan for Follow-up Strategies

Follow-ups are always crucial in any kind of promotion. For NetEase, it is strongly recommended that they segment their recruiting process, and conduct a brief survey each time a certain number of volunteers are on board. This way they would be able to best track where the volunteers come from. They could also collect data from websites where they have put up links and adjust their marketing plans.

III. A Review of RagBear’s Crowdsourcing Effort

RagBear is one of the largest fan-based volunteer translation group in China that mainly focus on translating subtitles of imported TV series. Their crowdsourcing workflow includes marketing to recruit, holding qualification tests, assigning tasks and finalizing the translation outputs.

Marketing Efforts

Being a fan-based group, RagBear doesn’t really need much marketing effort to attract volunteers. However, they still made efforts in recruiting, including watermark all their translated videos, posting volunteer information on them as rolling subtitles at the end of each episode, writing blogs and posts in social media, etc. Recently they’ve been cooperating with certain open course platforms in China such as NetEase to make their brand recognized by a larger audience. These efforts have proven to be successful, which can be measured by the constantly growing volunteer numbers displayed on their discussion forum.

Within the forum, there is a recruiting section. From the posts, one can tell that RagBear is not recruiting on a rolling basis, and there were times that the translator pool is full. These are the evidence that the methods are working for now.

Quality Assurance

The gate to being a volunteer translator in RagBear is the qualification test. I personally have taken one and noticed that in addition to enthusiasm and bilingual capabilities, it required basic knowledge of subtitling applications such as Ageisub.

Before taking the test the volunteer is required to choose the genre he/she is most interested in. Each genre is taken charge by one senior volunteer, who would review the test and decide which volunteers to recruit. The process may seem a bit arbitrary but it can be measured through their products that it is actually working well.

Motivating Volunteers

RagBear does reward the volunteers but the prize is nothing material. They get recognition in the forums and the chance to learn more advanced techniques in editing videos and subtitles. Moreover, they are given faster servers to download the original episodes and subtitles.

Conclusion

RagBear is one of the few fan-based subtitle translation group in China that is still operating. Many shut down because of censorship and poor product quality. However, since more and more Chinese are understanding English better, RagBear will have to address this issue by transforming its structure or consider cooperating with open course platforms to conduct educational subtitle translations. 

IV. Recommendations for LDS Church to Improve Translation Crowdsourcing Efforts

Based on the article Community Translation Efforts: Translation Department and Vineyard Partner Up and the presentation given in class, I observed the following challenges and problems that LDS Church is currently facing in their translation crowdsourcing effort and will provide my recommendations to address them. 

Resource Challenges:

It is stated in the article that there’s lack of volunteers from Chinese, Korean, Japanese, German, and Russian. Also, I noticed that currently there is a lack of motivation efforts to keep the volunteers going and making sure that they don’t abandon their projects halfway. In the presentation in class it is mentioned that competitions among volunteers are held sometimes, but from my perspective, it is not sufficient for motivating volunteers.

Recommendation:

For lack of volunteers, my suggestion is that LDS Church contact with Christian organizations in U.S. colleges and universities that has international students, for example, Christen Challenge at Washburn University. Such connections can be built via Facebook, YouTube, twitter or even the organizations’ websites.

With regard to motivating volunteers, there are plenty of things that LDS Church can do, such as giving out “swags”. Virtual rewards have been proven to be powerful incentives, such rewards could be rendering more voting powers in Lingotek, more leverage in proofreading translations, etc.

Quality Issues:

The common problems of translation crowdsourcing efforts are quality related, for example, the lack of consistency in terms of style and terminology. Currently, it seems very hard for LDS Church to make sure of the translation qualities because of the insufficiency in volunteers. However, they could also make adjustments or use Computer-assisted translation tools to tackle with quality issues.

Recommendation:

Within the Lingotek system, the voting process could be adjusted. Right now a translation needs only 2 votes to be “locked”, or considered done. Increasing the votes needed even by 1 would considerably raise the quality of translation and could also result in improvement of consistency.

Providing the translators with simple and clear stylesheets can also be useful in quality control. Furthermore, maintenance of terminology can be done simply by creating a google sheet where every volunteer can get access to and build the term base. Considering the number of volunteers and the cost of building up a more professional term base, a google sheet is more cost-efficient.

Feedbacks:

Another problem would be the fact that volunteers don’t have a place to communicate or share their problems and/or successes. In other words, when a problem occurs, a volunteer would not be able to immediacy reach out to other volunteers or someone in charge.

Recommendation:

This problem can be addressed by setting up a discussion forum in LDS Church website.Moreover, they could also make use of project management tools such as BaseCamp to get everyone that’s working on the same project on the same page. This way the workflow is well documented, the roles of each team member is clear, and it makes it a lot easier for volunteers to communicate.

V. Proposal & Presentation of Crowdsourcing Effort for China-Dolls Center for Rare Diseases 

Proposal for China Dolls

Presentation for China-Dolls

Introduction

The content of this portfolio is based on what I’ve learned during the Computer-Assisted Translation course at Middlebury Institute of International Studies at Monterey. We mainly used SDL Trados Studio and memoQ throughout the course, and after gradually getting familiar with these tools (with a little bit of struggle admittedly), I now understand why translators of our time should learn to work with them effectively as early as possible, as well as the pros and cons about machine translation (MT). My understanding is, the world is moving at an extremely high speed, and time and resources are limited for pure human translation. But translators shouldn’t be worried about being replaced by machines — they will always be a crucial part in the translation process. It’s the translators who refuse to use MT that will soon retire. A translation company not willing to use MT will not be cost-effective and scalable.

In this small portfolio I will mainly focus on some take aways on courses about post-editing MT.

Post-editing Machine Translation: Useful Tips

But first of all, why machine translation instead of human translation? We all know that with human assisted translation, we get high quality translations (well, most of the time), and they deal with unlimited topics. The problem is the cost, and more efforts will be needed to deal with different standards on definition of quality by different parties. With machine translation, although more times are needed to feed translating instruments, for contents that do not need high quality translations, it would be a win-win for both the vendor and the client. In my opinion, the key for success in implementing MT lies in whether the translators are willing to performance high quality post-editing with a relatively low price rate and with what standards. Here are a few tips that will help your post-editing team:

First of all, make sure that your translators are willing to work for you. Post-editing  MT translations often means lower rates, and more time pressure. The best practice would be to start with low discounts in order to ease the translators into the post-editing job. Second, with regard to the contents, because MT translations are expected to be “understandable” and “good enough”, you might find the following tips helpful:

• Any date formats are acceptable.
• Punctuation duplication is tolerable.
• Question marks are important, so if the source text is missing a question mark, always fix it.
• Quotation marks should be properly paired, quoting the correct material.
• Capitalizations can be remained as is.
• Acronyms and order of words don’t matter, as long as they can be comprehended without causing any confusion.
• Any added information should be deleted.
• Any missing information should be noted and added.

As you can see from the list above, post-editing is really about getting the content right and comprehendible, without worrying too much about the quality as your translation instrument had already taken care of most of it. Otherwise why would you spend efforts on feeding them?

 Sources and References:

https://www.linkedin.com/pulse/guide-post-editing-machine-translation-memsource-konstantin-dranch

http://www.nist.gov/itl/iad/mig/upload/BOLT_P3_PostEditingGuidelinesV1_3_3.pdf