Contract Disputes and Statutory Interpretation: The Linguist as Expert

The Oxford comma case, formally O'Connor v. Oakhurst Dairy decided by the First Circuit Court of Appeals in 2017, is the best-known recent example of how punctuation and syntactic structure create genuine legal ambiguity. Maine's overtime exemption statute listed several types of activity exempt from overtime protection, ending with 'the canning, processing, preserving, freezing, drying, marketing, storing, packing for shipment or distribution of agricultural produce, meat products, and seafood products'.

The question was whether the final item exempted 'packing for shipment or distribution' as one activity (packing, performed for either purpose) or 'packing for shipment' and separately 'distribution' as two distinct activities. The absence of a serial comma before 'or distribution' left both parsings grammatically plausible. Delivery drivers who distributed but did not pack argued the exemption did not cover pure distributors; the dairy argued it covered both.

The court sided with the drivers on grammatical grounds: the listformat of the preceding items suggested parallel structure, and the no-comma reading made distribution a standalone exempt activity that was grammatically inconsistent with the list's structure elsewhere. The case settled for approximately $5 million. A single punctuation mark had structured the entire dispute.

The systematic application of corpus linguistics to statutory interpretation in US courts is associated primarily with the work of law professor Stephen Mouritsen at Brigham Young University, whose 2012 article 'Hard Cases and Hard Data: Assessing Corpus Linguistics as an Empirical Path to Plain Meaning' (Columbia Science and Technology Law Review) and subsequent work with James Lee provided both the methodological framework and early case studies. Their approach is simple in outline: take the disputed statutory term, construct a corpus of text from roughly the same time as the statute's enactment, run a concordance analysis, and report what the distribution of usage shows.

Lockhart v. United States (2016) at the US Supreme Court provided an early high-profile application, though the majority opinion resolved the case without adopting corpus linguistics as the method. Justice Sotomayor's majority and Justice Kagan's dissent both discussed the ordinary meaning of 'related to' in a sex-offender sentencing provision; the exchange made visible both the promise and the complexity of corpus-based analysis. Neither justice used a formal corpus, but the decision's structure mirrored what a corpus analysis would do: asking how broadly 'related to' was used in comparable statutory and everyday contexts.

Several state supreme courts have gone further. The Utah Supreme Court in State v. Rasabout (2015) saw Associate Chief Justice Thomas Lee advocate corpus analysis for statutory interpretation in a notable concurrence, though the majority declined to adopt the methodology in that case. Michigan, Wisconsin, and Georgia state courts have cited corpus evidence in majority opinions. The US Supreme Court's decision in Niz-Chavez v. Garland (2021) saw Justice Gorsuch writing majority language that closely tracked corpus-based arguments about 'a notice' versus 'notice' in immigration law, though without formally calling it corpus linguistics.

Contract interpretation disputes arise when the parties hold different readings of their agreement. Under the plain-meaning rule that governs common-law contract interpretation, courts begin with what the text says in its ordinary sense and admit extrinsic evidence of intent only when the text is ambiguous. That threshold is precisely where linguistic analysis is applicable.

Lawrence Solan's work on contract and statutory language, particularly in 'The Language of Judges' (1993) and 'The Language of Statutes: Laws and Their Interpretation' (2010), documents the recurring patterns through which contractual language produces interpretive disputes: undefined terms that seem obvious but carry industry-specific meanings, sentences that are syntactically complete but referentially ambiguous (pronouns with multiple plausible antecedents), and scope clauses that use terms of degree ('material', 'substantial', 'reasonable') that are individually clear but unclear at the boundaries.

• Referential ambiguity: 'The company shall provide notice to the client at its registered address.' Whose registered address? The company's or the client's? Syntactic analysis and corpus comparison of similar clauses in comparable contracts can show which reading is normative for the contract type.
• Term-of-degree disputes: 'The contractor shall complete the work within a reasonable time.' Corpus analysis of 'reasonable time' in construction contracts shows the distributional centre of the phrase (typically measured in weeks for projects of a given size and type), giving courts an empirical baseline rather than an abstract appeal to what is 'reasonable'.
• Industry register versus ordinary meaning: A term that has a specific technical meaning in an industry may have been used by one party in that technical sense and by the other in its everyday sense. Expert evidence on register distribution, showing that in a particular industry corpus the term is used in the technical sense in 90 percent of instances, informs the ordinary-meaning analysis without deciding it.

The New Originalism, as developed by constitutional scholars over the past three decades, holds that the correct interpretation of a constitutional or statutory provision is the one an ordinary, competent speaker of English would have understood it to have at the time of adoption. This is a claim about historical public meaning, not about subjective legislative intent, and it converts a linguistic question (what did this word mean in 1789, or 1868, or 1954?) into an empirical one that corpus linguistics is suited to address.

The Corpus of Founding Era American English (COFEA), developed at BYU specifically for legal scholars, contains approximately 138 million words of American English text from the late 18th century. Analysing a constitutional term in COFEA gives a frequency distribution of its usage across legal, political, commercial, and everyday writing of the founding era. Justice Amy Coney Barrett, before her appointment to the Supreme Court, wrote about the methodological promise and limitations of such corpora in academic work.

The limitations are real and the method's leading practitioners acknowledge them. Historical corpora are uneven in their coverage of different text types, social registers, and geographic regions. The question of what an 'ordinary' speaker meant may have had no single answer in a linguistically diverse society. And the most contested constitutional questions often involve terms whose ordinary meaning in the founding era is precisely what both sides dispute, so the corpus data rarely resolves the legal question cleanly.

A responsible forensic linguist in a statutory or contract interpretation case maintains clear boundaries. The expert's remit is the empirical question: how was this term used? What does the syntactic structure of this clause support as its most natural reading? In which register and how frequently was this phrase deployed in comparable contexts? These are questions linguistics can address with the tools of the trade.

The legal questions are different. What did the legislature intend? Which party should win under the applicable interpretive canons? Is the technical or the ordinary meaning controlling given the context of this statute? These are legal and normative questions. A linguist who testifies to those questions has stepped outside expert competence. Courts have excluded or discounted corpus-based expert evidence when the expert presented frequency data as if it resolved the legal question rather than contributing evidence for the court to weigh.

• What linguistic expert evidence can show: frequency and distribution of a term's usage; which parsing is more syntactically natural; whether a claimed technical meaning is well-attested in the relevant register; whether a term had a different ordinary meaning at the time of drafting than it does today.
• What it cannot show: legislative or drafting intent; which interpretation the law should adopt; whether extrinsic evidence of purpose overrides the textual analysis; the legal consequence of the meaning finding.

Forensic linguists engaged in statutory or contract interpretation work generally follow a sequence that produces a defensible, transparent record of the analysis, one that opposing experts can scrutinise and courts can evaluate.

1. Define the question precisely
   State the specific disputed meaning precisely: is the question about a single word, a phrase, or the syntactic structure of a clause? Narrow questions produce more useful corpus evidence than broad ones.
2. Select and describe the corpus
   Choose a corpus that is temporally, geographically, and register-appropriate to the text. Document the corpus's composition, size, and known limitations. Historical texts require historical corpora; industry contracts require an industry corpus where possible.
3. Run the concordance and code the results
   Retrieve all instances of the disputed term or phrase and code each instance for which sense it reflects. Report inter-coder reliability if the coding required judgment. Publish the coding criteria so they can be examined.
4. Report frequency and context
   Present the distribution: what proportion of instances reflect sense A versus sense B? What collocates and contexts are associated with each sense? Acknowledge instances that are ambiguous or that do not fit either proposed sense cleanly.
5. State the conclusion within the appropriate scope
   Report what the corpus evidence shows about ordinary meaning distribution and stop there. Do not infer intent or decide the legal question. Make clear what the expert has measured and what lies beyond the measurement.