Trademark and Brand-Name Disputes: Linguistic Evidence

The legal test for confusion is consumer perception, not objective phonetic distance. This means a forensic linguist must assess how a real consumer, in a real purchase environment, would process the two marks, rather than how they look on paper side by side. The relevant modalities are sound (how they are spoken), visual appearance (how they are written or displayed), and meaning (what they connote).

• Phonological analysis: examines the number of shared phonemes, the positions of differences (initial consonants matter most, final consonants least), syllable count, and stress pattern. A two-syllable mark with initial stress will be processed differently from a four-syllable mark with secondary stress, even if they share four phonemes.
• Visual/orthographic analysis: assesses letter overlap, word length, and the visual gestalt of the mark in the trade dress context in which it appears. Consumers reading quickly in a retail environment often process approximate shapes rather than individual letters.
• Semantic analysis: considers whether two marks invoke the same concept or connotation, even when they look and sound different. 'Sunrise' and 'Dawn' for coffee brands may confuse on semantic grounds even with no phonological overlap.

The contested zone is in the middle: marks that are not identical but are not clearly different either. Linguists may use psycholinguistic priming experiments or consumer survey data to test whether the marks in fact produce confusion in the relevant population, rather than relying solely on theoretical phonetic distance metrics.

Genericness is the graveyard of brand names that became too successful. The classic examples document a pattern: a company coins a distinctive name for a new product category, the product succeeds, competitors and consumers start using the name as the category label rather than as a source identifier, and eventually the trademark loses its distinctiveness and protection. Aspirin (Bayer), escalator (Otis Elevator), zipper (B.F. Goodrich), and cellophane (Du Pont) all followed this arc.

The legal question is empirical: do consumers primarily understand the term as identifying a class of goods (generic) or as identifying a particular source (brand name)? Courts have used three types of evidence to answer this, and linguistics contributes to each.

1. Dictionary evidence
   If major dictionaries list the term as a common noun without a trademark symbol, that is some evidence of genericness, but weak because dictionaries lag usage and their editors may have accepted industry representations. A linguist can assess how the term appears in the actual citations supporting the dictionary entry.
2. Media usage analysis
   Searching news archives and web corpora for how journalists, bloggers, and other writers use the term tells a more current story. A term used consistently with a capital letter and in reference to one specific product is likely still a brand. The same term used in lower case, applied to products from multiple manufacturers, and in generic constructions ('a google for images', 'a jacuzzi in my bathroom') signals genericness.
3. Consumer surveys
   The 'Teflon survey' format, named after E.I. du Pont de Nemours and Co. v. Yoshida International Inc. (E.D.N.Y. 1975), tests whether consumers associate marks like 'Teflon' and 'Escalator' with one company or with a product type. Corpus linguistics has increasingly supplemented or replaced surveys because surveys are expensive, methodology-dependent, and susceptible to framing effects.

The marriage of corpus linguistics and trademark law began in earnest in US courts around 2010 and has accelerated since. The argument for using corpus data is compelling: rather than asking experts what a word means, you look at how actual speakers use it in the wild, in millions of unguarded contexts, without the coaching effect of a survey question.

For a genericness dispute, the analyst queries a large corpus, such as the Corpus of Contemporary American English (COCA, approximately 1 billion words), the News on the Web corpus (NOW, which exceeded 19 billion words by 2025 and grows continuously), or the Google Books Ngram Viewer, and retrieves all instances of the disputed term. The analysis examines:

• Capitalisation patterns: brand uses are typically capitalised; generic uses are lower case.
• Article patterns: 'a Xerox' or 'some Xerox' in constructions that imply the term is a count noun for a category rather than a proper name.
• Syntactic context: the term used as a verb ('google it', 'hoover the carpet') is strong evidence of generic use.
• Co-occurrence with brand-referencing language: proximity to competitor names or explicit source-identification ('the Google search engine' vs. 'a google') signals different functions.

The corpus linguist's job is to show what the evidence is, not what it means for the legal outcome. Interpreting the data's legal sufficiency is the court's task. This division of roles is what makes corpus evidence admissible as expert linguistics rather than as legal opinion.

The long-running contest over 'Budweiser' illustrates how linguistics, history, and jurisdictional variation interact in trademark law. The name comes from Budweis (now Ceske Budejovice) in Bohemia, where Budejovicky Budvar has brewed beer since 1895. Anheuser-Busch began using the same name in 1876, two decades earlier, and has sold beer under it in the United States for far longer.

The dispute is geographically fragmented. In the United States, Anheuser-Busch holds the trademark and Budvar's beer is sold as 'Czechvar'. In the United Kingdom, both once coexisted under the name 'Budweiser' after a 2012 Court of Appeal ruling found that both marks had sufficient historical use for coexistence. In Germany and Austria, Budvar has stronger protection. In Canada, Anheuser-Busch holds exclusive rights to the Budweiser name, and Budvar's beer is sold there as 'Czechvar'.

For linguistics, the case illustrates coexistence: two identical marks can coexist when consumers in a given market have learned to distinguish them by context, packaging, and price positioning. A forensic linguist asked to assess this might conduct a consumer study asking whether respondents recognise Budweiser as having multiple sources, or analyse corpus data for whether the name is consistently used with geographic or brand qualifying terms. The case also shows that the same empirical findings support opposite legal outcomes in different jurisdictions, a reminder that the expert supplies the linguistic facts and the court applies the law.

The forensic linguist in a trademark matter faces a particular pressure: the retaining party is usually a corporation with a clear financial interest in the outcome, and the analysis can seem to point one way or the other quite strongly. The expert's obligation is to the court, not to the outcome the instructing party wants, and that obligation becomes acute when the data is ambiguous.

• Corpus queries must be disclosed: the search terms, the corpus, the date of access, and the raw counts, so the opposing party can replicate or challenge the analysis.
• Survey methodology must be defensible: neutral question phrasing, appropriate respondent sampling, and published validation of the format used.
• Phonological analysis should be grounded in established phonetic metrics, not informal impressionism.
• The expert should state clearly what the analysis cannot determine, including the ultimate question of whether infringement occurred, which is a legal not a linguistic conclusion.