The Fight for Translation Memories

The Fight for Translation Memories

Translation memories are an irreplaceable component of any well-functioning, digitized language service provider (LSP), saving valuable time when translating repetitive documents that would otherwise take too long to painstakingly re-translate. Such memories are a valuable commodity, with LSPs and other language institutions buying and selling translation memories as the industry is expected to boom well into the decade, likely reaching $46.2 billion by 2027.


When certain phrases, paragraphs, or sentence-like “segments” are repeatedly used in a similar manner, they can efficiently be stored in an expansive database that is easily accessible to translators, thus operating as conveniently pre-saved translated “memories” – but what happens when these profitable memories are found in the hands of a single, highly powerful purchaser?


The Canada Translation Bureau, a formal department within the office of the Attorney General of Canada, amasses a record-breaking amount of translation memories as the single most purchaser of such memories in the world. Any corporate regulations that would typically apply to Canadian-based LSPs aren’t likely applicable to the CTB, allowing for this massive concentration of data to create a nearly unbreakable cycle of dependency upon the CTB for LSPs across the Great White North. The Translation Bureau operates as a gatekeeper for most translation memories, strategically buying low-quality translations for pennies on the dollar while reselling or “matching” them to LSPs at unfair rates, creating an expensive translation market in North America at the cost of well-researched memories. This places local Canadian translation firms at a clear disadvantage, with 25% of their revenues depleted due to this practice, according to legal research related to an active lawsuit against the CTB.


So, what’s next for the translation industry in Canada’s increasingly unaffordable market?


Legal Pushback – The CTB hasn’t achieved its current status as the premiere translation memory bank without legal trouble. Quattro, a leading Canadian LSP, filed a class action lawsuit in February of 2020 claiming, “it is forced to work from a Translation Memory that is flawed and polluted and that has not been revised according to standard practice”. If Quattro wins out, the legal ramifications could potentially require the Translation Bureau to alter its “weighted word clause”, which would upend the CTB’s current business model of paying less for words that are matched within its massive database of under-regulated translation memories.


Non-Canadian LSPs – If the Quattro case doesn’t pan out in favor of Canadian LSPs, companies seeking affordable, quality translations will likely go beyond Canada’s borders to meet their needs, a trend that has already begun since the Translation Bureau’s large-scale collection and control of translation memories started. LSPs in Latin America, for example, can offer more efficient, high-quality translations at a fraction of Canadian prices, possibly even beating out any potential changes in price should the CTB encounter more regulations if and when the class action lawsuit is settled.



Translation in Canada is essential, with the country espousing two official languages in navigating the demographic challenges of catering to both English and French speaking citizens. Usable translation memories are non-negotiable for business, yet what is being bought, produced, and sold by the nation’s official Translation Bureau has created a poorly curated monopoly on highly valuable content, placing Canadian LSPs at a clear disadvantage. Foreign LSPs are primed to fill in the gap, with the experience and price-point desired by firms looking to easily translate material for necessary audiences. Whether beating the Translation Bureau in court makes any dent in this trend remains to be seen, as the fight for translation memories has likely only just begun.


Although advancements in translating methods have had an impact, the presence of human touch is undeniably necessary, even if its role has changed.

Translation Memories – Much of what makes NMT an attractive option is its affordability and speed. In combination with a comprehensive translation policy set in place by an LSP, neuromachine translations can easily pick up on repeated phrases, words, or terminology, subsequently creating a database of translation memories. These memories eliminate the need for repeated translations, greatly reducing turn-around time and costs to the customer, especially when charged per word.

Terminology – Part of any respectable LSP’s approach to translation is understanding the nuances of the industry a client operates in. AI and machine learning can assist in this endeavor but fall short when considering contextual clues or fitting the tone of a particular audience. A good TMS can be fully put to use by creating a glossary of internal terms that a specific client uses and locking in such terms so as to avoid unintentional tampering by translators or any other individuals. This is critical to assisting translators working on a particular client, since many times a corporation may have an internal language that competitors or other companies within the same industry would hardly even recognize.

The common thread witnessed in the successful implementation of a TMS is a proven protocol that translators can easily follow and rely upon when encountering the more complex, nuanced aspects of translating. Sensitive information, repeated phrases, and company-specific lingo can all be handled smoothly and intelligently when combined with the experiential knowledge of translators who are adequately trained to fill in the less-obvious blanks.