Do you know the differences between Mandarin, Simplified Chinese and Traditional Chinese? Check out Mandarin or Chinese? Part 1

Drink deep, or taste not the Pierian spring – Alexander Pope

I had an interesting conversation with a client. In short, the client was questioning the cost of the 2 Chinese sections, primarily the translation. Honestly, I did not get what the buzz was, and I told him if it is possible to pass me more information via email. 10 minutes later, I ran into the office and saw the message. I was stunned, literally….

The message stated the cost of Traditional and Simplified Chinese is “incorrect”. Given to another vendor’s quote, the cost shall be 30% to 40% less.

“Not again!” I thought.

Most of sales and project managers, in reviewing the statement, may simply consider that the competitor simply undercuts the pricing for the job. However, this is not entirely true. The truth behind that quote may be:

  1. The competitor is using adaptation approach and launches the translations of both Traditional and Simplified Chinese as one (1) Chinese project.
  2. The competitor is able to utilize an open source character converter to produce Traditional Chinese translations from Simplified Chinese or vice versa.

In consideration of a single language vendor, the base word rate will drop. Therefore, the quoted total differs significantly in dollar value. That gets the client’s attention.

However, the problems in that quote are:

  1. Are the translators qualified to handle either Simplified Chinese or Traditional Chinese with an adaptation approach? Unlike Spanish and French variations, the 2 Chinese variations do not only have syntax and expression differences, but also character differences. Some of the Simplified Chinese characters are actually the Traditional Chinese older forms, which are obsolete in Traditional Chinese world. Furthermore, some converters cannot pick up these characters; therefore, editors may possibly skip them in the adaptation stage. 
  2. A localization project depends on translation memory to optimize leverage and achieve business efficiency and savings. Will the competitor be able to maintain a translation memory in both Chinese variations for future use? Will the client obtains it’s shared Intellectual Property (e.g. translation memory) without potential character corruptions?
  3. We talk a lot about continuous localization. The same concept is applicable. Assuming that Problem 2 is addressed, how would the competitor work out a business plan when the next content release hits? Will it be 2 different projects again? How can the client be absolutely confident that the same translated content in Chinese will not get reviewed twice and charged twice for review effort?
  4. How does the lower word rate justify the output quality? Is the rate at $0.xx/word a TEP rate included in the source locale of the adaptation (e.g. Simplified Chinese)? Is it just a translation rate and the quality of the source locale of the adaptation ensured by another round of linguistic QA? 

Reading the above, you may think that I absolutely object to the conversion solution. Well, you are wrong! (Finding my project manager’s hat now…)

I not only support this solution, but use it when:

I am 100% desperate to churn out both Chinese projects for deliveries in a relatively impossible timeline.

Alright, what will I do to avoid those problems above and ensure a quality output?

  1. Determine the priority locale between Simplified and Traditional Chinese with the client:
    • Since an adaptation is involved, a delay will occur. It is better to determine a priority local with the client at the scoping stage. The criteria may be the maximum cost-saving or the time to market.
    • A determined priority local is also helpful to the project manager in negotiations with the language service provider.
  2. Set expectations with the vendor who will provide source translation of the adaptation:
    • These expectations typically include daily translation throughput, translation constructs, accumulated glossary sheet, and a plan of independent linguistic review.
    • A emphasis on “syntax neutral” translation is a must because it will facilitate the adaptation throughput.
    • With the emphasis on “syntax neutral” translation, an independent linguistic review is required. It will flag at least some degree of local feasibility to the translation and potential quality issues so that the translation team can update them prior to post-production engineering stage.
    • Deliveries must be made in batches along with a glossary sheet in that batch delivery. This is important because batch deliveries enable the workflow and a glossary sheet in the batch facilitates the adaptation task with the search and replace function (if the term is translated differently in the adaptation locale).
  3. Set expectations with the localization engineering team to maintain assets:
    • Instead of maintaining the assets by single in-and-out traffic, we want the localization engineering team to work with the version control of these assets. Since multiple batches from the source translation will hit before finalization after the update of linguistic review, it is important that the engineering team and project manager track the versions of source translation deliveries in the event of project update in source content.
    • Instead of updating or creating a translation memory after translation finalized, we want to have translation memories tracked in line with the asset’s version control. This would benefit the translation teams later to weed out in-context (ICE) matches and facilitate productivity (if the tool supports Context TM function).
    • Find and test a character converter that optimizes the conversion. This is where the wonder takes place. There are many converters. The majority support the conversion very well. However, we still noticed some characters may not get converted due to the mapping. In addition, file format is also an issue. These conversion utilities support most MS files, .txt, .xml, and .html files. Sometimes, a change of filename suffix is required.
    • Use Context TM when an updated source hits. It is actually additional work when we work offline (i.e. in a CAT environment). Use this function prior to the conversion when the source file is prepared for the source translation of the adaptation (e.g. Simplified Chinese). Otherwise, use this function after the conversion is complete when we are preparing for the target locale of the adaption (e.g. Traditional Chinese).  
  4. Set expectations with the vendor who provides the target translation of the adaption:
    • Since the target translation of the adaption is in fact a review and editing process, the team shall expect, other than the glossary difference, the source translation may required a quality check and update.
    • To avoid an excessive check and update, we want the source translation team to provide “syntax neutral” translation so that the target translation team has minimum efforts to adjust the character-converted translation, except replacing the different glossary translations in the glossary sheet.
    • Review and correct the glossary translations for the target translation of the adaption prior to editing the work files. This would take a lot of efforts initially, but will benefit the team in a long run. If the team is technical enough (most unlikely), the team can actually work out a macro to automate the search and replace. (NOTE: I do not recommend the automation. It likely introduces false positive results and messes up the output.)

With all these adjustments in a workflow, we can start an adaption project for Simplified and Traditional Chinese to ensure almost all aspects of a localization project is covered. 

So, with all the extra work and necessary adjustments, why would I use this solution?

If you don’t have time to do it right, when will you have time to do it over? – John Wooden

The answer is quality, cost, and time.

Note the sequence! At the end, the “relatively impossible timeline” is no longer an impediment. If one looks at the requirements deeper, the quality is the ultimate impediment. Whether the source translation is Simplified or Traditional Chinese, translators like an average bilingual Joe on the street have stylistic preferences. They can go for wild rides in both Chinese translations separately and complete the translation at the same time. However, the quality reports from a separate linguistic review may differ. That means, if the manager is unfortunate, he/she may have triage sessions for potential duplicated issues with the 2 translation teams. Well, that is a waste of time to do it over. 

Instead, if the source translation team in the adaption approach carefully keeps syntax neutral in the translation stage, it at least gives the target translation team flexibility to edit the translations stylistically. On the other hand, an additional linguistic review* on the source translation will report linguistic issues for the reference of the target translation team.

*Linguistic review is typically based on a linguistic quality model (e.g. LISA). This model evaluates most of the aspects of the localization output, but here we are interested in the linguistic-related report. That way, the target translation team can verify if they may overlook some contents.

Alright, you may have a question, “Does this conversion thing really save money?”

Well, with this setup, I would say, “Yes, but marginal if we are running on the existing cost structure.” That is the reason I use it when I am desperate.

“Then, why do you even bother to write a few thousand words?” you may ask.

Imagine a Translation Management System (TMS) supports a conversion feature for Chinese locales. What will happen if this is true?

  1. The cost structure of zh-CN (or zh-TW) adaption can be standardized at word rate or post-editing rate, instead of hourly rate.
  2. The converter will be integrated, and the system can convert the source translation (e.g. Simplified Chinese) to the target translation (e.g. Traditional Chinese) real-time when the translator input a translation.
  3. No more tracking on translation memories! The system likely has translation memory sequence in place. When a job is created, the system leverages from the master translation memory and creates context match. Thus, the cost is reduced in the engineering, asset tracking, and translation memory tracking.

So, in a nutshell, if we look back at the conversation I had with the client, a cost difference of 30% – 40% is likely possible. However, the question is:

Do you want to do it right or do it over for either Simplified or Traditional Chinese?

This is the question, I think, for those innovative localization professionals to chew on….