Concept
Fact Versus Opinion
Defamation prima facie case: defamatory statement of and concerning the plaintiff
Fact Versus Opinion
Defamation law reaches only statements that can be wrong. A reviewer who calls a restaurant's decor tasteless risks nothing, because no evidence could show the remark false; a speaker who says the owner skims cash from the till risks a great deal, because the owner can prove the charge untrue. The first element of § 558, a false and defamatory statement concerning another, builds that difference into the tort.1 Restatement (Second) of Torts § 566 turns it into a working rule: a defamatory communication may take the form of an opinion, but the statement is actionable only if it implies the allegation of undisclosed defamatory facts as the basis for the opinion.2
The rule rests on what an audience can do with a judgment. When a speaker discloses the facts behind a conclusion, listeners can test the facts and decide for themselves whether the conclusion follows, and any damage to reputation comes from the facts, which stay open to proof of truth or falsity. A bare conclusion works differently. “In my opinion John Jones is a liar” implies knowledge of facts that lead to the conclusion that Jones told an untruth, and those implied facts can be false.3 The label therefore settles nothing: merely prefacing a defamatory statement with “in my opinion” does not shield the speaker from liability.4
Three subsections work out the line from different directions. The first applies § 566 to judgments that imply hidden facts. The second takes up the constitutional requirement that a statement on a matter of public concern be provable as false. The third deals with exaggeration, parody, and insult, which no reasonable audience takes literally and which therefore assert no fact that could turn out false.
Opinion Implying Facts
Section 566 has two parts, and each does its own work. The first refuses to let form decide the question, so a remark phrased as a judgment, such as calling someone a crook or a fraud, is tested like any other communication under the § 559 definition: does it tend so to harm the reputation of another as to lower him in the estimation of the community or to deter third persons from associating or dealing with him?5 The second part supplies the condition for liability, which is that the judgment imply an allegation of undisclosed defamatory facts.6 That condition ties the rule to § 558(a), because only a statement that asserts or implies facts can be checked against the world and found false.7
Consider what an audience does with a conclusion. When a writer states the facts and then draws an inference, readers hold the grounds and can judge the inference themselves, so an unfair comment does little harm if the facts are truthfully stated.8 When the writer offers a bare statement, or blends fact and comment so thoroughly that readers cannot separate inference from assertion, readers assume the damaging claim rests on grounds the writer knows and has not revealed. Those grounds are factual, they can be false, and the audience cannot test them. That is why “in my opinion” gives a speaker no cover: couching an accusation as opinion does not dispel the implication of facts behind it, and the statement can cause as much damage to reputation as the flat accusation.9
The Massachusetts decision in Nat'l Ass'n of Gov't Emps., Inc. v. Cent. Broad. Co. contrasts the two poles. A writer who says, without more, that a person is an alcoholic may have committed libel at first glance; a writer who says he saw the person take a martini at lunch and concludes from that sighting that the person is an alcoholic has disclosed the basis for the inference and stands differently.10 Beattie v. Fleet National Bank applies the same idea in a commercial setting. A bank's appraiser wrote a letter criticizing an appraisal Beattie had submitted and concluded that the report could be considered fraudulent. The Rhode Island Supreme Court held the letter not actionable, because the conclusion rested on disclosed, non-defamatory facts.11 Fraud is a serious word, yet every reader of the letter held the grounds and could decide whether the appraiser's inference was fair.
Milkovich v. Lorain Journal Co. supplies the opposite pole. A high school wrestling coach sued a newspaper and its columnist over a column implying that he had lied under oath in a judicial proceeding arising from a fight between two of his teams at a wrestling match. The Ohio courts treated the column as protected opinion. The United States Supreme Court held that the column's connotation of perjury was sufficiently factual to be proved true or false, and that the Ohio courts therefore erred.12
The Court's own example captures § 566. A speaker who says “In my opinion John Jones is a liar” implies knowledge of facts that lead to the conclusion that Jones told an untruth. Stating some supporting facts does not end the analysis either: if the facts offered are incorrect or incomplete, or the speaker's assessment of them is erroneous, the statement may still imply a false assertion of fact.13 Beattie and Milkovich differ in what the audience could check. The appraiser put his grounds on the table. The charge of lying under oath carried a factual connotation that could be proved true or false whatever words surrounded it.14
The Court refused to read the Gertz dictum that, under the First Amendment, there is no such thing as a false idea, as a wholesale exemption for anything labeled opinion. The dictum's marketplace-of-ideas origin points strongly to the view that the opinions it protected were the sort of thing discussion could correct, and a reading that covered every statement labeled opinion would ignore the fact that expressions of opinion may often imply an assertion of objective fact.19 In place of an opinion privilege, the Court found that existing doctrine already secures the breathing space free expression requires. Four safeguards do the work: statements on matters of public concern must be provable as false; statements that cannot reasonably be interpreted as stating actual facts are protected; public officials and figures must show knowledge of the false implications or reckless disregard of their truth, and private figures must show the fault Gertz requires; and enhanced appellate review checks the resulting determinations.20 The next two subsections take up the first two safeguards.
The implied-facts rule has a history that explains its shape. At common law a defamatory statement of opinion, if not privileged, was actionable the same as a statement of fact, and recent constitutional decisions have modified that rule.21 The common-law device for balancing vigorous public discourse against injury from invidious or irresponsible speech was the privilege of fair comment, which protected honest opinion on matters of public interest when it rested on a true or privileged statement of fact.22 Under the majority rule the privilege applied only to an expression of opinion and not to a false statement of fact, whether the fact was expressly stated or implied from an expression of opinion.23 Section 566 runs on the same logic. Comment on disclosed, true facts earns protection because readers can evaluate it, and an opinion that carries a hidden factual accusation loses protection because readers cannot. A student who stops at the word opinion misses the implied facts, and a student who treats every harsh conclusion as an accusation misses the work that disclosed facts do.
Provably False Statements
The first element of § 558(a) is a false and defamatory statement, and the First Amendment adds a filter that looks at the nature of the statement itself.26 In Milkovich v. Lorain Journal Co., the Supreme Court read Philadelphia Newspapers, Inc. v. Hepps to stand for the proposition that a statement on a matter of public concern must be provable as false before state defamation law can impose liability, at least where a media defendant is involved.27 Hepps itself held that the common-law presumption that defamatory speech is false cannot stand when a plaintiff seeks damages against a media defendant for speech of public concern.28 The two rules fit together. A rule that makes the plaintiff prove falsity assumes that evidence can establish falsity, and the provability requirement asks whether a statement is the kind of assertion on which evidence can speak at all.
The threshold reflects what a defamation jury can decide. A jury can find a statement false only by measuring it against facts in the world. When the statement is a judgment about ideas or policy that no record could confirm or refute, a verdict for the plaintiff would record the jury's reaction to the speaker's viewpoint, and liability would fall on whichever opinions the community disliked. Hepps observed that the burden of proof decides the outcome only when the evidence is ambiguous, and it resolved that doubt in favor of protecting true speech.29 A statement that no evidence could ever resolve presents the same doubt in its sharpest form, so the provability requirement removes it from liability at the outset.
Milkovich supplies both poles in one passage. “In my opinion Mayor Jones is a liar” can be actionable, while “In my opinion Mayor Jones shows his abysmal ignorance by accepting the teachings of Marx and Lenin” would not be, because the second contains no provably false factual connotation and so receives full constitutional protection.30 The second remark is a verdict on a person's ideas, and no document or witness could show it right or wrong. The first reports something the mayor did that either happened or did not.
The coach's own case shows the provable side. The Court held that the column's connotation of perjury was sufficiently factual to be proved true or false, and that this conclusion was not precluded by the fact that the statement was phrased as a question or appeared in the context of a public controversy.31 Whether a person lied under oath can be tested against transcripts and the witnesses who were present. In a separate suit over the same column, the Ohio Supreme Court had said that the charge, unlike a subjective assertion, was “an articulation of an objectively verifiable event.”32 The perjury charge therefore sits on the verifiable side of the line, and the Marx-and-Lenin remark sits on the other.
Provability depends on the kind of statement, not on the evidence a particular plaintiff happens to hold. The question is whether one could verify or falsify the statement by examining facts, so it does not ask whether this record contains enough to prove the statement false.33 The two questions are easy to run together because both involve evidence, yet they are answered at different points. Whether a statement is the kind of assertion evidence can address concerns the words and how a reader would take them. Whether the plaintiff can actually show falsity is the burden question that Hepps allocates, and it varies with which witnesses remain and which records survive. Treating provability as a question about the plaintiff's proof would wrongly shield accusations of concrete misconduct whenever the evidence has faded.
Courts commonly state the Milkovich inquiry as two questions: whether the statement is provable as false, and whether it can reasonably be interpreted as stating actual facts.34 The second question comes from the line of cases protecting statements that cannot reasonably be read as stating actual facts, and the next subsection takes it up.35 Either question can defeat a claim, so a statement must pass both before the plaintiff reaches the remaining elements.
Before Milkovich, lower courts had developed multi-factor tests for sorting fact from opinion, and the Ohio courts applied one in that very case, a four-factor approach of the kind exemplified by Ollman v. Evans. The factors were the specific language used, whether the statement is verifiable, the general context of the statement, and the broader context in which it appeared.36 The Ohio Supreme Court found that the first two factors favored treating the column as an actionable assertion of fact, but it let the other two prevail: the large caption “TD Says” would tell even the most gullible reader that the piece was opinion, and a sports page is a traditional haven for cajoling, invective, and hyperbole.37 Commentators group similar considerations as four elements that point toward fact: a core meaning understood by the intended audience; verifiability by objective tests; a context that increases the willingness to infer a factual statement; and a broader context that stresses fact or narrative, as opposed to editorials and reviews, where readers expect greater latitude from the author.38
The Supreme Court reversed. It declined to create a wholesale exemption for anything labeled opinion, and it held that a reasonable factfinder could conclude that the statements were accusations of perjury sufficiently factual to be proved false.39 The features the Ohio court treated as controlling did not change that answer. The language was not the loose, figurative, or hyperbolic sort that would negate the impression of a serious charge of perjury, the general tenor of the article did not negate it, and the sports-page setting did not negate the impression that the statements imply concrete facts.40 Context still bears on how a reasonable reader takes the words, but a caption or a section label cannot turn a verifiable accusation into protected opinion.
Falsity in this inquiry attaches to the facts a statement implies, and one statement can be false in two senses. “I think Jones lied” is false if the speaker did not really think so and false if Jones did not really lie.41 The second sense ordinarily supplies the basis for a defamation action, since it concerns facts about the plaintiff on which reputation depends, while the first may help establish malice where the plaintiff must prove it.
Clearing the provability threshold does not decide the case. After declining to create a separate privilege for opinion, the Court pointed to the safeguards that remain, among them the fault requirements: a public official or public figure must show that the false implications were published with knowledge of their falsity or reckless disregard of the truth, and a private figure suing over a matter of public concern must show the fault Gertz requires.42 The first Twitter defamation trial shows what follows the threshold. The singer Courtney Love was sued by her former lawyer over tweets that included a statement that the lawyer had been “bought off,” and the trial judge found unpersuasive her argument that the exaggeration common on the Internet made the tweet protected opinion. An accusation of being bought off describes an event that either happened or did not, so it clears the threshold. The judge also ruled that Love was a limited public figure, which required the lawyer to prove that Love either knew the statement was false or acted with reckless disregard of the truth, and the jury sided with Love.43
Hyperbole and Name-Calling
Milkovich's second question asks whether the words can reasonably be interpreted as stating actual facts, and the Court answered it against the columnist: the column was not the sort of loose, figurative, or hyperbolic language that would negate the impression that the writer was seriously maintaining that the coach had committed perjury.49 The Court drew on three earlier decisions that protect language a reasonable audience would take as exaggeration, insult, or invention: Greenbelt Cooperative Publishing Assn. v. Bresler, Letter Carriers v. Austin, and Hustler Magazine, Inc. v. Falwell.50 A statement that no reasonable listener would take literally fails the first element of § 558(a), because it asserts nothing that could be false.51
Two reasons support the carve-out. The first is the tort's own logic. Defamation compensates for the harm that follows when listeners come to believe something false and discrediting, and a reader who meets a vigorous epithet, a joke, or an obvious invention takes away the speaker's strong feeling and no belief that the plaintiff did the literal thing the words describe. The second is public debate, which the Court said should not suffer for lack of the “imaginative expression” or “rhetorical hyperbole” that has traditionally added much to the discourse of our Nation.52 A specific accusation of conduct that can be proved false stands differently, since it can mislead the audience, and the doctrine leaves it actionable.
Bresler supplies the model. Published reports quoted people who called a developer's negotiating position “blackmail,” and the developer sued for libel on the theory that the word accused him of a crime. The Court held that liability on that theory was impermissible. The reports were “accurate and full,” so “even the most careless reader must have perceived that the word was no more than rhetorical hyperbole, a vigorous epithet used by those who considered [the developer's] negotiating position extremely unreasonable.”53 Because the facts behind the label sat on the page beside it, the label carried no hidden accusation, and readers could weigh the bargaining for themselves.
The companion decisions apply the same question in other settings. In Letter Carriers, the use of the word “scab” in a union newsletter was not actionable, because it was used “in a loose, figurative sense” and was “merely rhetorical hyperbole, a lusty and imaginative expression of the contempt felt by union members.”54 In Falwell, a parody of an advertisement suggested that a well-known minister had an incestuous encounter with his mother in an outhouse, and the Court noted that the parody could not reasonably be interpreted as stating actual facts about him.55 Falwell arose under a state emotional distress claim rather than a defamation claim, yet the Court asked the same question about whether the words could reasonably be read as stating actual facts.56 Each decision identifies a different signal that told the audience not to read the words literally: the full and accurate report in Bresler, the union newsletter's expression of contempt in Letter Carriers, and the obvious invention of a parody in Falwell. An answer applying the doctrine should name the signal.
Because the test asks how a reasonable audience would take the words, the same word can state a fact in one setting and exaggerate in another. Justice Holmes observed in Towne v. Eisner that “[a] word is not a crystal, transparent and unchanged, it is the skin of a living thought and may vary greatly in color and content according to the circumstances and the time in which it is used.”57
The first paper put the facts beside the word, so readers weighed the word as a characterization of a bargaining position. A reader of the second story who accepts it has learned something that is either true or false, which is why the identical word lands on opposite sides of the line.
General insults sit at the far end of the same spectrum, because they carry no factual content that a record could confirm or refute. A South Dakota court held in Finck v. City of Tea that a statement that a police chief was a “dumb son of a bitch” was not actionable.58 The same question arises when a remark calls someone's son a “bastard”: the word may report the circumstances of his birth or simply abuse him, and abusive language with no factual content is generally treated as non-defamatory.59
Illinois applies the principle to commercial criticism. In Wilkow v. Forbes, Inc., a magazine article described a real estate entrepreneur's hard bargaining with lenders, and the court held that Illinois does not attach damages to name-calling. It pointed to Stevens v. Tillman, which collected examples such as “sleazy” and “rip-off,” and it observed that the entrepreneur's current and potential partners would have read the article as an endorsement of his strategy, because they want to invest with a general partner who drives the hardest possible bargain with lenders.60 A listener who hears that someone is sleazy learns what the speaker thinks of the plaintiff and nothing about any event, so no assertion exists that evidence could show to be false. That is the provability inquiry from the preceding subsection applied to a single word.
General accusations of racism show the line most clearly, because the word can function either as an insult or as an accusation of conduct. Defamation claims based on general allegations or insinuations that the plaintiff is a racist are likely to be rejected, while an allegation of specific racist conduct that can be proved false is a valid basis for a claim.61 In Schrottman v. Barnicle the conduct alleged was the use of a racial epithet, which either happened or did not.62 A label such as racist, sexist, or Uncle Tom is therefore likely to be dismissed as name-calling unless the accusation is made specific.63 A specific act is a fact that a witness could confirm or contradict, whereas a bare label hands the audience only the speaker's verdict.
Online speech brings the same analysis to a new medium. In Feld v. Conway, an online post said that the plaintiff was “crazy.” The court dismissed her defamation claim, reasoning that the post was made as part of a heated Internet debate over her responsibility for the disappearance of her horse and, read in that setting, could not reasonably be understood to state actual facts about her mental state, since it was obviously intended as criticism.64 The Internet setting does not make every post hyperbole, however. In the Courtney Love trial, the judge rejected the argument that a tweet saying the singer's lawyer had been “bought off” was opinion because exaggeration is common online.65 The two statements differ in what they say about their targets. “Crazy” thrown into a heated exchange is a general insult with no event behind it, while “bought off” describes a corrupt act that either occurred or did not, so it can be proved false. Passing this screen still left the claim undecided: the judge treated Love as a limited public figure, so the lawyer had to prove actual malice, and the jury sided with Love.66
The third step does the sorting, and it explains two opposite mistakes on an exam. One is to treat strong language as automatically protected, which overlooks that an accusation of a crime or a corrupt act is actionable when it can be proved false, as the coach's perjury charge in Milkovich was.69 The other is to treat a medium or a label as decisive, whether the label is opinion or the medium is the Internet, when the question is always how a reasonable audience would read these words in this setting. A statement that survives the screen still faces the remaining requirements, including falsity and the fault standard that fits the plaintiff, which the next sections develop.