I just came back from Localization World held in lovely Vancouver, BC. The primary audience at Localization World includes people involved in translation and localization. Most people fall into one of four categories:
- Localization Service Providers (aka translation companies)
- Localization tool vendors
- Employees of companies that are consumers of localization tools and services
With this as the primary audience, it is not surprising that topics are generally approached from the point of view of translation and localization.
This year, I spoke five different times over the course of three days. Three out of five of those sessions focused on terminology management. Terminology management is all the rage these days. Everyone is trying to bring some order to their growing set of unwieldy words.
As I participated in conversations about terminology, I realized that there is a huge divide in the thought processes around terminology management. There is also a large difference between how translators use tools and how content creators use tools. Let me explain.
The Terminology Database (Termbase)
Terminology management systems use a database to store terms and information about terms. We often refer to terminology databases as termbases. There are several terminology management systems on the market right now. These include MulitTerm, MultiCorpora, qTerm, and MultiTrans Prism, among others. They are all multilingual and focus on managing terms in multiple languages.
Basic Features of Termbases
All terminology management systems have ways to gather terms, validate terms, store terms, and allow you to include accompanying information, such as definitions and usage. Some have workflow processes built-in. The workflow processes include provisions for tasks such as:
- Gathering initial terms
- Suggesting additional terms
- Validating terms
- Approving terms
- Adding approved terms to the termbase
Uses and Control of Termbases
In almost 100% of the stories I heard, termbases are controlled exclusively by localization teams. This might be biased because of the people who attend Localization World. However, in my experience working with customers large and small, terminology management and terminology databases follow this same pattern. They are created and maintained almost exclusively by localization departments. On the surface, there is nothing inherently wrong with having localization groups control terminology. After all, who is responsible for dealing with more words than the localization team?
In addition, most localization groups have the linguistic expertise that makes them the most qualified to organize and take care of the corporate lexicon.
How Content Creation Groups Manage Terminology
Most content creation groups do not store terminology in a database. Instead, the tools of choice (or necessity) are either spreadsheets or tables. Terminology management from a content creation perspective is most often a manual process. The terms are gathered manually. The terms are entered into the spreadsheet or table manually. The terms are maintained manually. And the terms are looked up manually.
These manual processes are extremely cumbersome. Terminology managed via spreadsheet is almost always out of date. Who has time to work on the terminology list? Usually, the term list is largely ignored until so many years and product releases have gone by that someone in management realizes that the list is useless, and orders some poor soul to update it.
The Intersection of Content Creation and the Termbase
Providing the ability to look up terms in the termbase is touted by localization teams as the automated answer to terminology inconsistencies in the writing process.
Part of the reason for this is that terms are usually taken from the source content that is created by the content creators. This doesn’t mean that someone asks the documentation manager for the content. This means that as content moves through the workflow and lands in the localization group, localization people take the source content and grab terms out of it. Then, they put the terms into the termbase. Usually, the terminology management system includes an automated process to harvest terms from the content. Here is the typical terminology workflow:
- The localization group goes through quite a bit of time, work, and expense to create the termbase.
- The termbase is made available to content creators via a term look up or browsing interface. Usually, terminology management systems include some type of browser-based mechanism for viewing terms and their accompanying information.
- Depending on the software and the workflow, content creators may have the opportunity to suggest terms or term variants.
That Doesn’t Sound Too Bad
On the surface, allowing content creators to browse the terminology database sounds like a great idea. Having an automated term look up is a whole lot better than having to look up words in a spreadsheet. Sometimes just finding the spreadsheet is an enormous time-sink. In these instances, having a database look up is a godsend.
Providing the content creators with an opportunity to make suggestions for the termbase is also a great idea. The writers are really the people who know (or should know) which words are approved for use and which are not. They should also be the ones to use the terms correctly, particularly trademarks, service marks, and product names.
What’s So Bad About That?
While automated browsing sounds like a great idea, in reality, term browsing is ineffective and can be a waste of time and money for content creators. And here’s why.
There are two methodologies for governing the way content creators select and use terminology:
While term browsing is much easier than manual look up in a spreadsheet, it is ineffective because it is a look up process. In other words, it is nothing more than an automated pull process. And pull processes, even if they are automated, simply don’t work.
Pull – Push Tug of War
Here is the workflow for a pull methodology for term look up:
- The writer creates content.
- Someplace in the writing process, the writer needs to use a word that may or may not be managed – a term.
- The writer stops what he or she is doing to look up the word. The look up can either be manual via spreadsheet or table, or it can be automated via a term browsing function built-in to the termbase software.
- The word may or may not exist in either the spreadsheet or the termbase.
- If the word does exist, the writer reads the usage information and other information about the term.
- Where required, the writer changes how the term is used.
How long does this process take? Well, I’ve never scientifically measured it. But, it takes more than a few minutes. Regardless of how nice the term browsing interface is, no writer has time to do this. In fact, how many writers have the time to even contemplate whether a term should be looked up? The answer is zero. Or close to zero.
Like everyone else in your company, writers have deadlines. The deadlines are almost always unreasonable, even under the best circumstances. In an Agile environment, writers barely have time to type. They certainly don’t have time to go traipsing through a list of words, automated or manual.
Pull methodologies do not work because they are disruptive and time-consuming.
A push methodology provides the terminology information to the writer, when the writer needs it. It requires specialized software. Here is the workflow:
- The writer creates content.
- At some point in the writing process or at the end of the writing process, the writer clicks a button to check the terminology.
- The software pushes the information to the content writer. In other words, the software looks at each word in the document and compares it to the termbase. If there is a mistake, for example the writer accidentally used an outdated product name, the software flags the incorrect term and suggests the accurate term.
Push methodologies work because they are instantaneous and do not require the writer to remember or look up anything.
Push methodology is the mechanism used by computer-aided translation software. As the translator processes content, the CAT software automatically pushes information to the translator. The software checks each segment (or phrase) for a match in the translation memory database. If translation exists, it appears on the screen while the translator is working on the content. Again, this push methodology works because it is instantaneous and does not require the translator to do a manual “pull” lookup.
The Fallacy that Surrounds Termbase Look Up
The bottom line is that localization teams do not think about writing. If they did, they would see how futile it is to provider writers with term browsing and lookup capabilities. Sure, it is a whole lot better than looking something up in a difficult-to-find spreadsheet. Unfortunately, it is still a look up. And even a well-intentioned content creator doesn’t have time for that.
We’re here to help with all of your terminology needs. Just click here and we’ll be in touch.