Plural of “corpus”

The plural of corpus is corpora.

Why

"Corpus" is Latin for "body," a third-declension neuter noun whose plural is "corpora" — an entirely different pattern from "focus," a fourth-declension masculine noun that pluralizes as "foci." English borrowed "corpus" mainly for scholarly contexts: a body of writing or evidence (a legal corpus, a corpus of literature), and more recently, in linguistics, a large structured collection of text used for language research (a "text corpus").

Because "corpus" has stayed almost entirely within academic and legal registers rather than drifting into everyday conversation the way "focus" has, a regularized "corpuses" has essentially no foothold anywhere in professional usage — the word simply hasn't been used casually often enough for a simpler plural to take hold the way it has for more common Latin borrowings.

Usage note

Use "corpora" without hesitation in linguistics, law, and literary scholarship; it's the essentially unchallenged standard form. "Corpuses" does appear occasionally in more informal writing about bodies of work, but it remains the clear minority choice even there.

In corpus linguistics specifically, "corpora" is a foundational technical term — researchers build and compare corpora of millions or billions of words to study how language is actually used, and getting the plural wrong in this field would be a noticeable lapse. Digital humanities has expanded the word's reach considerably in recent years, with entire academic conferences now devoted to comparing corpora built from historical newspapers, social media, or literary archives, and this rapidly growing field has, if anything, reinforced "corpora" as the standard plural rather than introducing pressure toward a simpler alternative, since the term's users are overwhelmingly specialists who expect the traditional form.

Common real mistakes

Writers unfamiliar with Latin declensions sometimes assume "corpus" follows the same -i pattern as "cactus" or "focus" and mistakenly write "corpi" — this is incorrect; "corpus" is a different declension entirely, and its only standard plural is "corpora."

The wider pattern

"Corpus" and "genus" share a grammatical family: both are Latin third-declension neuter nouns, which is why both take an -a-based plural ("corpora," "genera") rather than the -i ending seen in second-declension words like "cactus" and "focus" — a genuine grammatical distinction, not an arbitrary difference between similar-looking words. Legal scholars use "corpus" in the specific phrase "corpus delicti," the body of facts proving a crime occurred, and while this legal phrase is typically discussed in the singular for any one case, comparative legal scholarship examining multiple cases' corpora of evidence still relies on the standard Latin plural.

Examples

• The linguists compiled two corpora, one spoken and one written, for comparison.
• Her dissertation draws on several corpora of 19th-century personal letters.
• The legal team assembled a corpus of evidence spanning a decade.
• Researchers built separate corpora for each regional dialect they studied.
• The scholar's life work produced two major corpora of translated poetry.
• Digital humanities projects now compare corpora across dozens of languages.
• The study measured word frequency across five distinct corpora.
• Investigators treated the leaked documents as a single corpus of evidence.
• Two overlapping corpora were merged to create a larger training dataset.
• The archive holds several corpora of correspondence from the same era.
• The AI model was trained on several massive text corpora.
• Two separate corpora were used to compare formal and casual writing styles.
• The archive holds corpora of letters spanning three generations of the family.
• Linguists cross-referenced corpora from five different English-speaking regions.
• The prosecution built its case around a thin corpus of physical evidence.
• Scholars compared corpora of trial records from two different centuries.
• The research team assembled corpora specifically for machine translation training.
• Two rival corpora of ancient inscriptions offered conflicting historical accounts.
• The team is building corpora specifically for underrepresented languages.
• Two overlapping corpora revealed inconsistencies in the historical record.