Portia Boulger filed an action for defamation and invasion of privacy based upon a tweet by James Woods, a well-known movie actor and producer. The district court denied Woods's motion for summary judgment, due to Boulger's lack of service, but granted his motion for judgment on the pleadings, finding that the tweet was not an actionable statement of fact under Ohio law. Boulger now appeals the grant of the motion for judgment on the pleadings as to her defamation claim, and Woods cross-appeals the denial of his motion for summary judgment. For the reasons that follow, we affirm.

I. BACKGROUND

On March 11, 2016, then-presidential candidate Donald Trump held a rally in Chicago, Illinois. That evening, the Chicago Tribune newspaper posted a photograph on its Twitter account of a woman at the rally, wearing a Trump T-shirt, and giving a Nazi salute-a salute with her right hand raised straight up in the air. On March 12, 2016, Twitter user @voxday posted the Nazi salute photograph, together with a photograph of Portia Boulger and a caption identifying Boulger as an "Organizer (Women for Bernie)." (Def. Mot. for J. on the Pleadings, R. 7, PageID 61.) The two photographs and caption were accompanied by the (false) statement, "The 'Trump Nazi' is Portia Boulger, who runs *475 the Women for Bernie Sanders Twitter account. It's another media plant." ( Id .) Shortly thereafter, Woods tweeted the same two pictures, along with a short biography of Boulger, and added: "So-called #Trump 'Nazi' is a #BernieSanders agitator/operative?" (Comp., R. 1, PageID 3.) At the time, Woods had more than 350,000 followers on Twitter.

That same day, March 12, 2016, multiple news outlets identified the woman in the Nazi salute photograph as Birgitt Peterson, a Trump supporter residing in Yorkville, Illinois. Woods did not delete his original tweet, but instead tweeted a follow-up: "Various followers have stated that the Nazi Salute individual and the #Bernie campaign woman are NOT the same person. #Chicago #Trump." (Def. Mot. for J. on the Pleadings, R. 7, PageID 62.)

On March 22, 2016, counsel for Boulger wrote to Woods's attorney, asking that Woods delete his tweet and issue a retraction and apology. Woods deleted the tweet the next day, and posted three new tweets:

1. "I have an opportunity to clarify something I challenged immediately when it hit Twitter. Portia A. Boulger was NOT the 'Nazi salute lady.' "

2. "Ms. Boulder [sic] has reached out to me and asked me to use my many followers to stop people from harassing her. I am more than happy to do so."

3. "Though she supports @BernieSanders, I am happy to defend her from abuse. I only wish his supporters would do the same for other candidates."

(Comp., R. 1, PageID 5.)

In the eleven days between Woods's initial tweet and the tweet's deletion, Boulger "received hundreds of obscene and threatening messages, including death threats," as well as numerous telephone calls. ( Id. at 5-6.) Boulger stated that due to Woods's tweet, she suffered "severe emotional distress including sleeplessness, episodes of reasonable apprehension of personal assault or attack, anxiety and depression." ( Id. at 8.)

Boulger subsequently filed the instant action against Woods on March 3, 2017, alleging defamation and invasion of privacy under Ohio law. On June 1, 2017, Boulger filed a motion for extension of time to complete service of process on Woods, arguing that, despite her best efforts, she had been unable to serve Woods properly. The district court granted Boulger's motion, extending the service deadline to August 7, 2017.

Despite the extension, Woods filed an answer to the complaint on June 7, 2017, asserting, inter alia , insufficient service of process and lack of personal jurisdiction (due to the lack of service). The same day, Woods also filed a motion for judgment on the pleadings, arguing that Boulger's claim for defamation failed as a matter of law because the tweet at issue was a question and not an actionable statement of fact. The next month, the parties filed a joint report under Federal Rule of Civil Procedure 26(f), in which they recommended that "discovery be stayed pending a Decision on the Motion for Judgment on the Pleadings." ( Rule 26(f) Report, R. 12, PageID 127.) Woods also noted in the Rule 26(f) report that he "contests personal jurisdiction" and "has not been served." ( Id. )

Boulger still failed to serve Woods timely. Accordingly, on August 15, 2017, Woods filed a motion for summary judgment, or in the alternative, motion for dismissal, due to Boulger's failure to perfect service. The district court found, however, that Woods waived his jurisdictional defenses through his conduct and therefore denied his motion for summary judgment. But the *476 court granted Woods's motion for judgment on the pleadings, finding that Woods's tweet could be interpreted as a question and not a statement of fact, and that the tweet was protected under Ohio's innocent construction rule.

Boulger now appeals the decision on the merits, and Woods cross-appeals the jurisdictional issue.

II. ANALYSIS

A. Jurisdiction

We must first decide whether we have jurisdiction to review the merits of Boulger's claim. In the absence of "proper service of process, consent, waiver, or forfeiture, a court may not exercise personal jurisdiction over a named defendant." King v. Taylor , 694 F.3d 650 , 655 (6th Cir. 2012) (citations omitted). And without personal jurisdiction, a federal court is "powerless to proceed to an adjudication." Id. (quoting Ruhrgas AG v. Marathon Oil Co. , 526 U.S. 574 , 584, 119 S.Ct. 1563 , 143 L.Ed.2d 760 (1999) ).

It is undisputed that Woods was never properly served. In the district court, Boulger sought to excuse her service failure, arguing that Woods waived his defenses in two ways: (1) by failing to include the defenses in his motion for judgment on the pleadings and (2) through his conduct. The district court rejected Boulger's first argument, but found that Woods had waived his ability to challenge service through his conduct.

We review a district court's ruling on waiver for an abuse of discretion. King , 694 F.3d at 659 . "An abuse of discretion occurs if the district court relies on clearly erroneous findings of fact, applies the wrong legal standard, misapplies the correct legal standard when reaching a conclusion, or makes a clear error of judgment." Id. at 660 (quoting In re Whirlpool Corp. Front-Loading Washer Prods. Liability Litig. , 678 F.3d 409 , 416 (6th Cir. 2012) ).

1. Waiver Through Failure to Include Defenses in Rule 12 Motion

Under the Federal Rules of Civil Procedure, a defendant who files a motion under Rule 12 and fails to raise the defense of insufficient service of process "waives" that defense. King , 694 F.3d at 656 (citing Fed. R. Civ. P. 12(b)(1), (g)(2), (h)(1)(A) ). 1 A defendant cannot be deemed to have waived defenses, however, that "were not ... available at the time they could first have been made." Holzsager v. Valley Hosp. , 646 F.2d 792 , 796 (2d Cir. 1981).

While Woods challenged the sufficiency of service in his answer, he acknowledges that he failed to include the defense in his motion for judgment on the pleadings. But as Woods notes, he filed his motion for judgment on the pleadings in June 2017, two months prior to the service deadline. Accordingly, any argument made by Woods in June 2017 regarding insufficient service of process would have been premature. See King , 694 F.3d at 661 ("[A] motion to dismiss on the basis of improper service made during the period for service may properly be denied as premature."). The district court was therefore correct in holding that Woods did not waive his insufficient *477 service of process and lack of personal jurisdiction defenses by failing to include them in his motion for judgment on the pleadings because, at the time of filing, the defenses were not yet available.

2. Waiver Through Conduct

"Even where a defendant properly preserves a Rule 12(b) defense by including it in an answer"-as Woods did here-"he may forfeit the right to seek a ruling on the defense at a later juncture through his conduct during the litigation." King , 694 F.3d at 658 (footnote omitted) (citing Hamilton v. Atlas Turner, Inc. , 197 F.3d 58 , 60 (2d Cir. 1999) ). Put simply, a defendant's appearances, filings, and actions in the district court may constitute "legal submission to the jurisdiction of [that] court." Gerber v. Riordan , 649 F.3d 514 , 519 (6th Cir. 2011) (quoting Days Inns Worldwide, Inc. v. Patel , 445 F.3d 899 , 905 (6th Cir. 2006) ). But not all conduct serves as constructive consent to personal jurisdiction. Instead, courts must ask whether a defendant's conduct "has given the plaintiff 'a reasonable expectation' that the defendant will defend the suit on the merits or whether the defendant has caused the court to 'go to some effort that would be wasted if personal jurisdiction is later found lacking.' " King , 694 F.3d at 659 (quoting Gerber , 649 F.3d at 519 ).

"Determining what constitutes waiver by conduct is more [an] art than a science ... and there is no bright line rule." State Auto Ins. Co. v. Thomas Landscaping & Constr., Inc. , No. 2:09-cv-735, 2011 WL 3475376 , at *6 (S.D. Ohio Aug. 9, 2011) (quoting Pruco Life Ins. Co. v. Wilmington Trust Co. , 616 F.Supp.2d 210 , 216 (D.R.I. 2009), aff'd in relevant part , 494 F. App'x 550 (6th Cir. 2012) ). We note though that "it is relatively easier to find forfeiture of a service defense[,]" as opposed to a personal-jurisdiction defense. King , 694 F.3d 650 (noting "that service of process is simply the means by which a defendant receives notice of an action and is formally brought within a court's jurisdiction, whereas personal jurisdiction concerns the fairness of requiring a defendant to appear and defend in a distant forum"); see also 5B Charles A. Wright & Arthur R. Miller, Federal Practice & Procedure § 1353, at 338 (3d ed. 2004) ("Although the questions of personal jurisdiction and service of process are closely interrelated, service of process is merely the means by which a federal court gives notice to the defendant and asserts jurisdiction over him."). Ultimately, we must consider all of the relevant circumstances in determining whether waiver by conduct has occurred. King , 694 F.3d at 659 . And our analysis must be deferential to the district court's assessment of the situation. Id.

Turning to Woods's conduct, one filing "stands out as showing the obvious intent of the defendant[ ] to submit to the personal jurisdiction of the Court." Gerber , 649 F.3d at 525 (Moore, J., concurring). As the district court noted, although Woods raised the defenses of insufficient service of process and lack of personal jurisdiction in his answer, he immediately filed a motion for judgment on the pleadings in which the defenses were not included. The motion for judgment on the pleadings was filed several months early-because Woods had not yet been served-and necessarily sought a decision on the merits. Woods's motion was thus "inconsistent with the idea that the district court lacked personal jurisdiction over the defendant[ ]." Id. Indeed, Woods's actions demonstrate that he sought to have the district court use its power over the parties to reach a decision on the merits, and required the court to expend significant efforts in doing so. See Parchman v. SLM Corp. , 896 F.3d 728 , 734 ("[T]he voluntary *478 use of certain district court procedures serve[s] as constructive consent to the personal jurisdiction of the district court ...." (alterations in original) (quoting Gerber , 649 F.3d at 519 )). Such voluntary participation in the litigation gave Boulger "a reasonable expectation that [Woods would] defend the suit on the merits." King , 694 F.3d at 660-61 (quoting Gerber , 649 F.3d at 519 ).

The filing of the motion for judgment on the pleadings therefore created a reasonable expectation that Woods would defend the suit on the merits. Any other holding would create a perverse outcome. One can imagine a litigant asking the court to proceed on the merits, and then, only if the court's decision is unfavorable, seeking to re-assert jurisdictional defenses. The district court thus reasonably concluded that Woods's actions constituted waiver. And, importantly, that kind of decision is precisely the discretionary judgment that district courts are empowered to make in these kinds of circumstances. In another case, the district court might reasonably conclude that the appearance of gamesmanship was lower than that here and reach a different decision. But the risk is certainly there, and we will not disturb the discretion of the district court's judgment in this case. We therefore proceed to the merits.

B. Defamation

"We review a district court's judgment on the pleadings 'using the same de novo standard of review employed for a motion to dismiss under Rule 12(b)(6).' " Fed. Deposit Ins. Corp. v. Amfin Fin. Corp. , 757 F.3d 530 , 533 (6th Cir. 2014) (quoting Tucker Middleburg-Legacy Place , 539 F.3d 545 , 549 (6th Cir. 2008) ).

"Under Ohio law, 'it is for the court to decide as a matter of law whether certain statements alleged to be defamatory are actionable or not.' " Susan B. Anthony List v. Driehaus , 779 F.3d 628 , 632 (6th Cir. 2015) (quoting Yeager v. Local Union 20 , 6 Ohio St.3d 369 , 453 N.E.2d 666 , 669 (1983), overruled on other grounds by Welling v. Weinfeld , 113 Ohio St.3d 464 , 866 N.E.2d 1051 (2007) ). In order to establish a claim for defamation, a plaintiff "must show (1) that a false statement of fact was made, (2) that the statement was defamatory, (3) that the statement was published, (4) that the plaintiff suffered injury as a proximate result of the publication, and (5) that the defendant acted with the requisite degree of fault in publishing the statement." Driehaus , 779 F.3d at 632-33 (quoting Am. Chem. Soc. v. Leadscope, Inc. , 133 Ohio St.3d 366 , 978 N.E.2d 832 , 852 (Ohio 2012) (internal citation omitted)). Failure to establish any one element is fatal to a defamation claim. Id.

Here, only the first element is at issue. The case thus boils down to one question: does Woods's tweet constitute a false statement of fact? See Scott v. News-Herald , 25 Ohio St.3d 243 , 496 N.E.2d 699 , 705 (1986) ("Expressions of opinion[,]" as opposed to statements of fact, "are generally accorded absolute immunity from liability under the First Amendment."). The Ohio Supreme Court has held that in determining "whether a statement constitutes protected opinion or actionable fact, courts should consider the totality of the circumstances, including factors such as: (1) the specific language used; (2) whether the statement is verifiable; (3) the general context of the statement; and (4) the broader context in which the statement appeared." Bentkowski v. Scene Mag. , 637 F.3d 689 , 693-94 (6th Cir. 2011) (internal quotation marks omitted) (citing Vail v. The Plain Dealer Publ'g Co. , 72 Ohio St.3d 279 , 649 N.E.2d 182 , 185-86 (1995) ). Each factor is to be addressed, "but the weight given to any one will conceivably vary *479 depending on the circumstances presented." Vail , 649 N.E.2d at 185 .

This case, however, differs slightly from those cases contemplated by the Ohio Supreme Court. Here, rather than determining whether a statement is protected opinion or actionable fact, we are asked to analyze whether Woods's question constitutes a statement at all. But Ohio case law indicates that even in a unique circumstance such as this one, the four-prong, totality-of-the-circumstances test is still the appropriate framework. Indeed, in McKimm v. Ohio Elections Commission , 89 Ohio St.3d 139 , 729 N.E.2d 364 , 371 (2000), the Ohio Supreme Court was asked to determine whether a cartoon could be construed as an implied statement of fact. In its analysis, the court noted the four-factor test and explained that "[a]ll four factors of Ohio's test for distinguishing a statement of fact from an opinion depend on the reasonable reader's perception of the statement-not on the perception of the publisher." 729 N.E.2d at 371 . While the court's analysis in McKimm did not explicitly apply the four-prong test, the Ohio Supreme Court made clear, and has continued to make clear, that the four-prong test is the appropriate framework in analyzing whether a message can be construed as a statement of fact. See Wampler v. Higgins , 93 Ohio St.3d 111 , 752 N.E.2d 962 , 971 (2001) (explanatory parenthetical states that the court in McKimm relied on the four-prong test). We are thus satisfied that the four-prong test should be utilized in determining whether Woods's tweet constituted a statement of fact under Ohio law.

1. Factor 1-Specific Language Used

Under the first factor, courts seek "to determine whether the allegedly defamatory statement has a precise meaning and thus is likely to give rise to clear factual implications." Bentkowski , 637 F.3d at 694 (quoting Wampler v. Higgins , 93 Ohio St.3d 111 , 752 N.E.2d 962 , 978 (2001) ). Woods argues that because the specific language at issue was a question-and thus not likely to give rise to clear factual implications-his tweet was not a statement of fact. In the alternative, Woods argues briefly that his tweet was not actionable because the word "Nazi" is "inherently imprecise and subject to myriad subject interpretations." (Appellee Br. 12 (quoting Wampler , 752 N.E.2d at 979 ).) However, the cases Woods relies upon-involving statements such as "surgery Nazi" and "Little Hitler"-are easily distinguishable from the language used in this case. Consequently, this argument has no merit.

Woods spends more time arguing that his use of a question mark inherently precludes his tweet from qualifying as a statement of fact:

Basic grammar establishes that "[a] question mark follows every question for which an answer is expected." BRYAN A. GARNER, GARNER'S MODERN ENGLISH USAGE 752 (4th ed. 2016) (emphasis added). The question mark signaled to Mr. Woods' Twitter followers his "lack of definitive knowledge about the issue," and also "invites the readers to consider" various possibilities. Partington v. Bugliosi , 56 F.3d 1147 , 1157 (9th Cir. 1995). From grade school on, question marks at the end of a sentence signal the sentence is a question. Applying those same basic grammar rules, requires the dismissal of the Complaint.

(Appellee Br. 11 (emphasis in original).)

Although this court has never confronted the issue of whether questions can (or cannot) be defamatory, Ohio courts have. In Schoedler v. Motometer Gauge & Equip. Corp. , the Ohio Supreme Court *480 held that "[a] mere insinuation is as actionable as a positive assertion, if the meaning is plain, and it has been held repeatedly that the putting of the words in the form of a question will in no w[ay] reduce the liability of the defendant." 134 Ohio St. 78 , 15 N.E.2d 958 , 961 (1938) (internal citations and quotation marks omitted). We agree, as it is easy to see how commentators and journalists could use questions "as tools to raise doubts ... about a person's activities or character while simultaneously avoiding defamation liability." Abbas v. Foreign Policy Grp., LLC , 783 F.3d 1328 , 1338 (D.C. Cir. 2015). For example, "a question's wording or tone or context sometimes may be read as implying the writer's answer to that question." Id. (emphasis in original); see also Bryan A. Garner, The Chicago Guide to Grammar, Usage, and Punctuation (2016) ("[A] rhetorical question is phrased in the interrogative structure but is meant as an emphatic or evocative statement .") (emphasis added). For this reason, a blanket protection of defamation liability for any and all questions is inappropriate.

It is worth noting, however, that other circuits have opined that "it is generally settled as a matter of defamation law ... that a question, 'however embarrassing or unpleasant to its subject, is not accusation.' " Abbas , 783 F.3d at 1338 (quoting Chapin v. Knight-Ridder, Inc. , 993 F.2d 1087 , 1094 (4th Cir. 1993) ); see also Bugliosi , 56 F.3d at 1157 (holding that a defendant's use of a question mark indicates "lack of definitive knowledge about the issue"). 2 Even still, these holdings do not amount to a per se rule that questions cannot be defamatory.

As to the question in this case, on March 12, 2016, Woods posted two photos-one photo of a woman giving the Nazi salute and one of Boulger. These photos had been posted by other Twitter users who affirmatively identified Boulger as the woman giving the Nazi salute, yet Woods chose to frame his tweet as a question. According to Woods, the question mark illustrates that he intended the tweet to ask a question to his followers. But Woods's intentions are irrelevant; all that matters to this court is the reasonable reader's interpretation of his tweet. See McKimm , 729 N.E.2d at 371 .

Some readers likely viewed the tweet as an insinuation that the woman in the Nazi salute photograph was Boulger. It seems equally plausible, though, that other readers interpreted the tweet as posing a question. It would thus be nearly impossible to say that Woods's tweet had a precise meaning as required by Ohio law. See Wampler , 752 N.E.2d at 979 ; see also Abbas , 783 F.3d at 1338 ("There is no good or predictable way to neatly divide (i) the questions that are routinely posed in America's robust public forums from (ii) the kinds of questions that would be actionable as defamation by implication ...."). With this ambiguity in mind, the first factor weighs in favor of nonactionability.

2. Factor 2-Verifiability

The second factor considers whether the allegedly defamatory statement *481 is verifiable. "Where the ... statement lacks a plausible method of verification, a reasonable reader will not believe that the statement has specific factual content." Vail , 649 N.E.2d at 186 (alternation in original) (citations and internal quotation marks omitted). "A statement is deemed verifiable if: (1) the author represents that he has knowledge or evidence that substantiates the statements, and (2) there is a plausible method to verify the statements." SPX Corp. v. Doe , 253 F.Supp.2d 974 , 980-81 (N.D. Ohio 2003) (citing Scott , 496 N.E.2d at 706-07 ; Vail , 649 N.E.2d at 186 ; Wampler , 752 N.E.2d at 979 ). For example, a non-verifiable statement would be one that refers to an individual's "internal motivation," because no plausible method to confirm the veracity exists. See Bentkowski , 637 F.3d at 694 ; see also Ferreri v. Plain Dealer Publ'g Co. , 142 Ohio App.3d 629 , 756 N.E.2d 712 , 721 (2001) (noting that "[t]here are no objective tests to establish whether someone's conduct is 'scandalous,' whether an individual is 'beyond the call of reason,' has the 'qualities that make a mere lawyer into a good jurist,' or 'cares a good deal more about himself and his image' than children"). On the other hand, Ohio courts have held that "an accusation of perjury was 'an articulation of an objectively verifiable event' that could be proven 'with evidence adduced from the transcripts and witnesses present at the hearing.' " Wampler , 752 N.E.2d at 979 (quoting Scott , 496 N.E.2d at 707 ).

The district court held that a "question itself cannot be proven or disproven because questions, by their nature, lack truth values." (Op. & Order, R. 21, PageID 207.) But as we have already noted, a question's wording, tone, or context can be read as implying a writer is asserting a statement of fact. Take a rhetorical question as an example. A person could walk outside, see that it is snowing, and tweet "Is it seriously snowing again?" Read in context, the question can be reasonably interpreted as implying a statement of fact that it is snowing, and such a statement is undoubtedly verifiable. Accordingly, a general rule that all questions cannot be verified is inappropriate.

This case, of course, does not present so obvious an example. We therefore proceed to analyzing the context of Woods's tweet.

3. Factors 3 and 4 -General and Broader Context of the Statement

Consistent with this court's previous approach, the third and fourth factor-general context and broader context-are analyzed together. See Bentkowski , 637 F.3d at 695 ("We evaluate both factors together for ease of discussion, and use general context to refer to the entire work at issue, and broader context to refer to the publication in which the work appears.").

The general context is analyzed to determine "the larger objective and subjective context of the statement[,]" so that the alleged defamatory statements are not examined in isolation. Id. (quoting Scott , 496 N.E.2d at 707 ). Put another way, the court considers the "immediate context" in which the allegedly defamatory statement appears. Wampler , 752 N.E.2d at 980 ; see also id. ("[T]he language surrounding the averred defamatory remarks may place the reasonable reader on notice that what is being read is the opinion of the writer.").

The district court concluded that "because the nature of a 'tweet' is fundamentally different from a statement appearing in the context of a longer written work," and because most Twitter users do not sit down and read a Twitter account in chronological order, there is no 'general context' to examine. (Op. & Order, R. 21, PageID 209.) While we agree that a tweet *482 "does not map neatly" into our prior analysis of general context, that does not end the inquiry. Instead, we consider-as both parties suggest-the allegedly defamatory statement in the context of Woods's other tweets on March 12, 2016.

A review of Wood's Twitter feed from March 12, 2016, shows that although he posted news articles, his tweets were frequently accompanied by his own colorful commentary. For example, Woods retweeted a USA Today article regarding protesters at Donald Trump's Chicago rally, along with his own caption: "Trump blames 'thugs' for cancelling Chicago rally ... Rubio, Cruz, Kasich, Clinton blame #Trump. That says something." (R. 9-2. PageID 87.) Woods also retweeted an MSNBC article regarding the same protesters with the caption: "Bernie Sanders supporters shut down a Trump rally in Chicago | Try this kind of assault in Arizona... #dangerous." ( Id. at PageID 88.) Finally, Woods tweeted a picture of a man walking on an American flag, and stated "Why #Trump is going win [sic]..." ( Id. at PageID 87). These tweets illustrate that a reasonable reader of Woods's tweets on March 12, 2016, likely knew that he made frequent use of sarcasm, exaggeration, and hyperbole-characteristics more likely seen in an opinion, rather than a statement of fact. See Scott , 496 N.E.2d at 708 . Thus, the general context could lead a reasonable reader to believe the tweet at issue was not a statement of fact.

Turning to the broader context, "the Ohio Supreme Court has recognized that '[d]ifferent types of writing have ... widely varying social conventions which signal to the reader the likelihood of a statement[ ] being either fact or opinion.' " Bentkowski , 637 F.3d at 695 (first and second alterations in original) (quoting Scott , 496 N.E.2d at 708 ). "Statements appearing in such locations as forum and commentary newspaper sections, or other venues often associated with 'cajoling, invective, and hyperbole,' are more likely opinion." SPX Corp. , 253 F.Supp.2d at 981 (quoting Scott , 496 N.E.2d at 708 ). Courts must examine the type of medium at issue and consider how it would influence the reader's viewpoint on whether a message is one of fact or opinion. See id.

Twitter is a medium for users to express both opinions and disseminate news. For example, a Twitter user who tweets his or her thoughts on various celebrities is an account that is more analogous to an editorial section of a newspaper. Cf. Vail , 649 N.E.2d at 185-86 (finding that a column that appeared on the Forum page of the newspaper and titled "Commentary" gave a reader the message that the column would convey the personal opinion of the writer, as distinguished from a news story). But the Twitter account of an online news source, such as the New York Times, is not meaningfully distinguishable from a hard copy news story. Consequently, it is clear that Twitter can be used to disseminate both factual accounts and assertions, as well as commentary and opinion. We therefore cannot say that this factor weighs in favor of either party.

At base, allegedly defamatory statements will ultimately fall somewhere on the spectrum "between paradigmatic statements of fact (such as 'Mr. Jones had ten drinks at his office party and sideswiped two vehicles on his way home') and paradigmatic statements of opinion (such as 'Mr. Jones is a despicable politician')." Wampler , 752 N.E.2d at 977 (citing Ollman v. Evans , 750 F.2d 970 , 978 (D.C. Cir. 1984) ). Here, the factors do not definitively tip the scale in either direction. Accordingly, we turn to Ohio's innocent construction rule.

*483 4. Innocent Construction Rule

As both parties acknowledge, the tweet at issue must also be analyzed, in the alternative, under Ohio's innocent construction rule. See New Olde Vill. Jewelers, Inc. v. Outlet Comm., Inc. , 202 F.3d 269 (6th Cir. 2000) (table opinion) ("Ohio follows the innocent construction rule in adjudging defamatory statements."). Under the rule, "a statement reasonably susceptible to both a defamatory and an innocent meaning must be construed, as a matter of law, to have an innocent meaning." Id. "It matters not that the defamatory meaning is the more obvious one. So long as the statement may reasonably be read to have an innocent meaning, the innocent construction rule commands that the statement be deemed non-defamatory." Id. (citations omitted).

Here, the tweet at issue is reasonably susceptible to both a defamatory meaning-that Woods was asserting Boulger was the woman giving the Nazi salute-and an innocent meaning-that Woods was merely asking his followers a question. Because Woods's tweet could reasonably be read to have an innocent meaning, under the innocent construction rule the tweet, as a matter of law, is not actionable.

III. CONCLUSION

For those reasons, we affirm the judgment of the district court.

NALBANDIAN, Circuit Judge, concurring in part and concurring in judgment.

CONCURRING IN PART AND IN THE JUDGMENT

I agree with the majority that James Woods cannot be liable for defamation because his tweet is most reasonably understood as posing a genuine question. But I would not apply Ohio's four-pronged, totality-of-the-circumstances test to reach this conclusion. That test, which the Ohio Supreme Court adopted to distinguish between statements of fact and opinion, fits awkwardly here. I write separately to explain why.

Ohio adopted the totality-of-the-circumstances test to resolve a specific problem. Unlike its federal counterpart, Ohio's constitution provides absolute immunity from liability for expressions of opinion. See Vail v. The Plain Dealer Publ'g Co. , 72 Ohio St.3d 279 , 649 N.E.2d 182 , 185 (1995) ; Milkovich v. Lorain Journal Co. , 497 U.S. 1 , 21, 110 S.Ct. 2695 , 111 L.Ed.2d 1 (1990). Two cases helped establish that rule. First, in Scott v. News-Herald , 25 Ohio St.3d 243 , 496 N.E.2d 699 (1986), the Ohio Supreme Court unequivocally (and it turns out, incorrectly) held that the First Amendment shields opinion speech from suit for defamation. Id . at 705. When the United States Supreme Court disagreed in Milkovich , Ohio reaffirmed the rule under its own constitution. See Vail , 649 N.E.2d at 185 . The Ohio constitution provides a "separate and independent guarantee of protection for opinion," the court held. Id . Because of that, courts presiding over defamation suits must determine whether-as a matter of law-the offending speech falls within the protected boundary of opinion.

To do that, the Ohio Supreme Court adopted the four-pronged, totality-of-the-circumstances test the majority relies on today. See Scott , 496 N.E.2d at 706 ; Vail , 649 N.E.2d at 185 . That test is "to be used when determining whether a statement is fact or opinion." Vail , 649 N.E.2d at 185 . Each prong provides a different parameter for making that assessment, and it is within this fact/opinion paradigm that courts have continued to use it. See, e.g., Wampler v. Higgins , 93 Ohio St.3d 111 , 752 N.E.2d 962 , 971 (2001) (explaining that *484 statements of opinion are protected if they pass " Scott 's four-part test"); see also Bentkowski v. Scene Magazine , 637 F.3d 689 , 693-94 (6th Cir. 2011) (citing the four factors "[t]o determine whether a statement constitutes protected opinion or actionable fact").

The hitch, of course, is that our case is not about whether the tweet was a statement of fact or opinion-it's about whether the tweet was a statement at all. The defendant argues that he cannot be liable for defamation because he asked a genuine question. And a question, he says, cannot be defamatory because the first element of the tort requires proving the defendant made a false statement of fact. See Am. Chem. Soc'y v. Leadscope, Inc. , 133 Ohio St.3d 366 , 978 N.E.2d 832 , 852 (2012). So our task is to determine whether we ought to construe an ostensible question as implying an actionable statement of fact. That's a different problem than sorting out fact from opinion, and it is one that a mechanical application of the four-pronged test seems ill-suited for.

The problem with applying the four-pronged test presents itself most starkly when trying to assess the second factor, verifiability. A statement is verifiable when it is "objectively capable of proof or disproof." Wampler , 752 N.E.2d at 979 . This prong helps courts distinguish between facts and opinions because an opinion, lacking any plausible method of verification, cannot be proven or disproven. See Vail , 649 N.E.2d at 186 . So an accusation of perjury, for example, looks factual, because one can prove or disprove perjury. But an accusation of heartlessness looks like an opinion, because it lacks any objective measure of truth. See Wampler , 752 N.E.2d at 979-80 .

How, though, could we apply that test in any meaningful way to a case like this, where the speaker posed a question? Asking whether a question is verifiable is like asking how tall an elephant weighs. A question is an inquiry , not an assertion, so it's neither verifiable nor unverifiable. To be sure, it might be answered , or it might imply some kind of assertion-both of which could be verifiable. But a genuine question lacks a truth value, so asking about verifiability only highlights how the Ohio Supreme Court designed the totality test to solve a different problem than the one here.

The majority suggests that we should stick with the four-pronged approach because some questions might be verifiable if the tone, wording, or context imply that the writer asserted a fact. Take for example, a rhetorical question like, "Is it seriously snowing again?" That question, the majority correctly explains, implies the assertion that it is snowing outside-an indisputably verifiable claim. But what the majority misses is that verifiability does nothing to solve the antecedent problem, which is determining whether the question actually implies an assertion of fact. And that's the only issue in our case.

To make this more concrete, consider a different question, similar in tone, wording, and context as the one offered by the majority: "Is the professor seriously droning on again?" Just like the example used by the majority, this question reasonably implies an assertion-that the professor is "droning on." But unlike the first example, the assertion is not verifiable because the phrase "droning on" is a "standardless statement[ ] not amenable to objective proof or disproof." See Wampler , 752 N.E.2d at 979-80 . It is quintessential, unverifiable opinion. Yet the question impliedly asserts a statement for all the same reasons as the majority's question about snow. Verifiability, in other words, is irrelevant. It does nothing to help courts determine *485 whether a question should be read as implying an assertion.

Rather than try to awkwardly work around this issue, I would resolve this case in a much simpler way by asking whether a reasonable reader would interpret the language as a genuine question. In McKimm v. Ohio Elections Commission , 89 Ohio St.3d 139 , 729 N.E.2d 364 (2000), the Ohio Supreme Court applied a reasonable-reader standard to determine whether a cartoon contained an implied factual assertion. Instead of going through the four-pronged totality test, the court simply considered whether a reasonable reader would believe that the cartoon implied an assertion of fact. Id. at 371-72 . 1 While not directly on all fours, the issue in McKimm resembles the one here. Both cases circle around the same problem. Just as the McKimm court asked whether a reasonable reader would interpret the cartoon as implying an assertion, I would resolve this case by asking how a reasonable reader would interpret the offending tweet.

This approach also aligns with the only Ohio case addressing whether questions can be defamatory, Schoedler v. Motometer Gauge & Equipment Corp. , 134 Ohio St. 78 , 15 N.E.2d 958 (1938). There, the court held that "[a] mere insinuation is as actionable as a positive assertion, if the meaning is plain." Id. at 961 . Asking whether the "meaning is plain" looks a lot like asking how a reasonable reader would interpret the language. If a reasonable reader could interpret the tweet as posing a genuine question, rather than implying an assertion of fact, the meaning is not "plain."

Applying this standard, I would affirm the district court because a reasonable reader would interpret the tweet as posing a genuine question. An ordinary person sees a question mark and assumes the writer is asking a question. That does not make the punctuation definitive, but it sets a baseline. And beyond a bit of awkward syntax, there is no basis to conclude that a reasonable reader would find that the question mark simply obscured a defamatory accusation. The question, for example, is not rhetorical. Woods posed the question and provided two different pictures that the reader could look at to provide an answer, suggesting he meant it as a genuine inquiry. Nor is this the kind of sentence that implies a secondary accusation in the form of a question, such as, "How long have you been stealing from the company?" That kind of question might be actionable, but it's not at issue here. So I would affirm the district court's decision as a matter of law because a reasonable reader would not interpret the tweet as an implied statement of fact. 2

Admittedly, the reasonable-reader standard overlaps significantly with the four-pronged approach adopted by the majority. How a reasonable reader might interpret *486 a question surely takes into account things like words and context. But the standard we use matters, even if in practice it often amounts to "a distinction without a difference." Vail , 649 N.E.2d at 185 . The four-pronged totality test was designed and adopted to separate fact from opinion. Since that's not the issue here, I would not employ the test to resolve it.