• Research suggests dictionary and media usage not effective to prove genericide
  • Counsel urged to push back if dictionary evidence cited in cases
  • Suggests digital language databases more beneficial, but urges caution

A recent study has analysed the three most common forms of evidence used in genericism cases, and found that corpus linguistics – ie, the use of a digital language databases – could prove to be “more beneficial” for rights holders looking to protect their brand from the threat of genericide. However, the author urges cautions over its use, saying that courts should reconsider the use of linguistic data altogether and reclaim the primary significance test in genericism cases.

The research, entitled Reclaiming the Primary Significance Test: Dictionaries, Corpus Linguistics, and Trademark Genericide, was penned by Neal Hoopes, an associate at Covington & Burling and a former executive director at Legally Mine. Talking to World Trademark Review, Hoopes said he was inspired to look at the topic of genericism after reading one of the most famous trademark genericide cases, Murphy Door Bed Co v. Interior Sleep Systems, Inc. “I was taken aback by the lack of analysis in the court's opinion regarding genericide and dictionaries,” he explained. “As I read other genericide cases, I found that many courts misunderstand the role that linguistic evidence (dictionaries and media usage) may objectively play in determining the genericness of a trademark. I decided to conduct my own study on the effectiveness of corpus linguistics in determining trademark genericide. I wanted to clarify the theory behind genericide to give guidance on how courts may use linguistic evidence objectively.”

To that end, the study focuses on the most used sources for evidence to determine whether a trademark has become generic. Looking at the most used evidence in genericism cases, dictionaries, the study bluntly suggests they “cannot prove” that a well-known trademark has become generic because it “cannot and does not” claim to reveal the primary significance of a word. “Dictionaries are useful only to determine the range of possible meanings of a trademark,” Hoopes writes. “Thus, a dictionary may reasonably demonstrate that some segment of the public uses the mark generically in some circumstances. But using dictionaries to prove broader claims, such as the primary significance of a term, gravely misunderstands how dictionaries are compiled and organised.”

In fact, in further comments to us, Hoopes recommends brand owners push back if dictionary evidence is raised in court: “Dictionary entries are not strong indicators of trademark genericide – when a rights holder is embroiled in litigation and the challenger points to generic uses in a dictionary, a trademark owner should force the court to confront the weaknesses inherent in using dictionaries to determine the primary significance of a term, as required by the Lanham Act.”

Turning to another commonly used source of genericism evidence, media usage, the study suggests that using examples of a brand term from newspapers or magazines is “slightly more useful” than dictionaries. However, due to the way that parties present media usage evidence, it states that this method is still “methodologically unsound”, expanding: “In the end, isolated media uses verify only that a generic sense to a trademarked term is linguistically possible – that some media writers or editors use the term generically some of the time. But proving a word sense linguistically permissible does not demonstrate the majority usage of a term.” Going further, it notes: “Courts must determine the primary significance of a term, but isolated examples, even ‘numerous examples’ from newspapers and magazines, cannot prove much more than dictionary definitions – that the trademarked term has a genuine generic sense. But it is a non sequitur to conclude that since isolated generic usages exist, they must comprise the majority of uses.”

Therefore, the most useful tool that Hoopes found for evidence in a genericism case is that of corpus linguistics – specifically, computer databases composed of naturally occurring words used in context. For example, the Corpus of Contemporary American English (COCA) is a database of more than 530 million words, equally divided among spoken, fiction, popular magazines, newspapers, and academic texts. There are two main advantages that a language database has over dictionaries and media usage, the study finds; firstly, a search for a term returns a random sampling, and so avoids claims of “cherry-picking favourable examples”, and secondly, words in the corpus are tagged for grammatical content, such as verb versus noun usage.

Indeed, Coopes used the COCA to analyse various famous brands that have been threatened with genericide in the past. For example, when looking at Xerox reference in the COCA, 75% of sampled uses were specifically referencing the Xerox Corporation, while just 7% of mentions were generic references. On the flip-side, only 3% of Band-Aid uses were used in the trademarked sense; 15% of uses were “clearly generic usages” and the other 82% of uses were “ambiguous” references (eg, “a small trickle of blood seeps from under his Band-Aid”). The results therefore allow a somewhat objective figure of usage that can be presented in court.

Nonetheless, despite the advantages, Coopes suggests that “corpus linguistics fails to be the effective substitute for dictionaries and media usage that some trademark litigators might hope”. He expanded on this point to us by claiming: “Rights holders should be wary of a court drawing improper inferences from corpus linguistics data. Corpus linguistics can be a powerful tool for trademark owners, especially for well-established trademarks. But rights holders should be attentive to how challengers use this data, ensuring that the court does not equate public usage with primary significance.”

So while the study suggests corpus linguistics could “prove more beneficial” than other tools for analysing genericide evidence (especially for companies seeking to protect their trademarks), it has its limits – especially in conclusively proving a term’s genericness. To overcome this, Coopes concludes that it may be down to the courts: “The Lanham Act requires a showing of primary significance. Courts should maintain majority usage and primary significance as distinct concepts and, in this way, should reclaim the primary significance test.”