**533Two sets of petitioners have challenged the Attorney General's certified ballot title for Initiative Petition 28 (IP 28). We review the ballot title for substantial compliance with ORS 250.035(2). See ORS 250.085(5) (stating standard of review). For the reasons explained below, we conclude that the ballot title does not substantially comply with ORS 250.035(2) and refer the ballot title to the Attorney General for modification.
Article I, section 8, of the Oregon Constitution prohibits laws "restraining the free expression of opinion, or restricting the right to speak, write, or print free on any subject whatever." See State v. Robertson , 293 Or. 402, 649 P.2d 569 (1982) (interpreting Article I, section 8 ). This court held in Vannatta v. Keisling , 324 Or. 514, 931 P.2d 770 (1997) ( Vannatta I ), that making contributions to candidates is protected expression and that laws limiting the amount of contributions that a person, corporation, or union makes to candidates or political committees violate Article I, section 8.324 Or. at 537-39, 931 P.2d 770 ; see Vannatta v. Oregon Government Ethics Comm. , 347 Or. 449, 222 P.3d 1077 (2009) (clarifying Vannatta I ).
IP 28, if adopted, would add an exception to the constitutional protections recognized in Vannatta I . More specifically, IP 28 would add the following sentence to Article I, section 8 :
"Laws consistent with the freedom of speech guarantee of the United States Constitution may regulate contributions and expenditures, of any type or description, to influence the outcome of any election; provided, that such laws are adopted or amended by initiative or by an elected legislative body by a three-fourths vote."
A law will come within the exception set out in IP 28 if three conditions are met: the law must (1) be "consistent with the freedom of speech guarantee of the United States Constitution"; (2) "regulate contributions and expenditures * * * [made] to influence the outcome of any election"; and (3) be "adopted or amended by initiative or by an elected legislative body by a three-fourths vote."
**534The Attorney General certified the following ballot title for IP 28:
"Amends Constitution: Allows laws regulating contributions/expenditures *969made to influence elections; Measure 47 (from 2006) becomes law
"Result of 'Yes' Vote: 'Yes' vote amends constitution, allows laws regulating contributions and expenditures to influence election outcomes; triggers Measure 47 (from 2006), regulating campaign expenditures, requiring certain disclosures.
"Result of 'No' Vote: 'No' vote retains Oregon Constitution's existing free-expression provision; Measure 47 (from 2006), regulating campaign expenditures, requiring certain disclosures, remains inoperative.
"Summary: Amends Constitution. The Oregon Supreme Court has interpreted the Oregon Constitution's free-expression provision ( Article I, section 8 ) to prohibit limits on many political campaign contributions and expenditures. The proposed measure amends Article I, section 8 to allow laws that regulate contributions/expenditures to influence elections. Such laws must be adopted or amended by initiative or by three-fourths vote by legislative body. If amendment passes, Measure 47 (enacted by voters in 2006) will become law and will alter Oregon's current election laws. Measure 47 limits campaign contributions/expenditures, establishes new reporting/advertising disclosure requirements for the sources and amounts of campaign contributions/expenditures. Laws regulating campaign contributions and expenditures (including Measure 47) still must comply with the First Amendment to the United States Constitution."
Petitioners challenge the caption, the "yes" and "no" result statements, and the summary. We begin with their challenges to the caption.
ORS 250.035(2)(a) provides that a ballot title must contain a "caption of not more than 15 words that reasonably identifies the subject matter of the state measure." The "subject matter" of a ballot title is "the 'actual major effect' of a measure or, if the measure has more than one major effect, all such effects (to the limit of the available words)." Lavey v. Kroger , 350 Or. 559, 563, 258 P.3d 1194 (2011) (citation **535omitted). To identify the "actual major effect" of a measure, we consider the "changes that the proposed measure would enact in the context of existing law." Rasmussen v. Kroger , 350 Or. 281, 285, 253 P.3d 1031 (2011).
Petitioners challenge the caption on primarily three grounds. First, they argue that the caption should not mention Measure 47 (2006) because the subject of IP 28 is broader than its effect on a specific law and, more importantly, that effect is at best speculative. Second, they contend that the caption should make clear that the extent to which IP 28, if adopted, would "regulate" campaign contributions and expenditures is unclear. Finally, they argue that the caption should mention that a law may regulate contributions and expenditures only if it is passed by initiative or by an elected legislative body by a three-fourths vote.
Before turning to petitioners' objection to the reference in the caption to Measure 47 (2006), we first discuss briefly the history of that measure. See Hazell v. Brown , 352 Or. 455, 461-63, 287 P.3d 1079 (2012) (setting out that history). As noted above, in 1997, this court held that Article I, section 8, prohibits statutes limiting the amount of campaign contributions made by persons, corporations, and unions. Vannatta I , 324 Or. at 541, 931 P.2d 770. The court also recognized, as the parties conceded, that campaign expenditures on behalf of a candidate or measure were protected expression under Article I, section 8. Id. at 542, 931 P.2d 770.
Nine years later, in 2006, two initiated ballot measures were submitted to the voters. Hazell , 352 Or. at 461-63, 287 P.3d 1079. Measure 46 (2006) was a state constitutional amendment that would have "expressly allow[ed] laws limiting or prohibiting election contributions and expenditures." Id. at 462, 287 P.3d 1079. Measure 47 (2006) was a statute that, among other things, limited the amount of contributions that could be made to candidates and also the amount of expenditures that directly supported or opposed a candidate or a party. Id. at 462-63, 287 P.3d 1079. A majority of voters approved the statutory but not the constitutional measure, with the result that the voters enacted a statute that, in whole or in part, was unconstitutional *970under Article I, section 8, as interpreted in Vannatta I. **536One section in Measure 47 (2006) anticipated that possibility. It provided:
" 'If, on the effective date of this Act, the Oregon Constitution does not allow limitations on political campaign contributions or expenditures, this Act shall nevertheless be codified and become effective at the time that the Oregon Constitution is found to allow, or is amended to allow, such limitations.' "
Hazell , 352 Or. at 464, 287 P.3d 1079 (quoting section 9(f) of Measure 47 (2006) ). In Hazell , this court explained that the "limitations cited in section 9(f) are most logically viewed broadly as the same kind of limitations struck down in Vannatta I * * *." Id. at 466, 287 P.3d 1079.1 Accordingly, the court held that, as a result of section 9(f), Measure 47 (2006) would be codified but would not become effective until "the Oregon Constitution is found to allow, or is amended to allow, such limitations." Id. at 464, 287 P.3d 1079 ; see id. at 470, 287 P.3d 1079.
With that background in mind, we turn to petitioners' objections to the statement in the caption that "Measure 47 (from 2006) becomes law." On that issue, we agree with petitioners that the subject or "major effect" of IP 28 is to amend Article I, section 8, to permit a class of laws-laws that regulate campaign contributions and expenditures. The question whether IP 28 would revive a specific law enacted some 12 years earlier is neither the subject nor the major effect of the measure. Beyond that, as we explain below in discussing petitioners' challenges to the summary, the Attorney General's statement that Measure 47 (2006) "becomes law" is inaccurate because the effect that IP 28 will have on Measure 47 (2006) is at best speculative. Finally, we agree with one of the amici that the unexplained reference to a 12-year old measure in the caption, even if otherwise appropriate, is more likely to confuse the voters than to enlighten them. For those reasons, the caption should not refer to Measure 47 (2006).
Petitioners Markley and Hedbor raise a second objection to the caption. They argue that its use of the word **537"regulate" fails to alert voters that the extent to which IP 28 would permit regulation of campaign expenditures and contributions is far from clear. As we understand Markley and Hedbor's argument, they contend that "regulations" of campaign contributions and expenditures can range from disclosure and reporting requirements to substantive limitations on the permissible amount of contributions and expenditures. See Hazell , 352 Or. at 465, 287 P.3d 1079 (distinguishing those differing types of regulations). In a similar vein, this court has explained that laws that regulate the secondary effects of a medium of speech differ from laws that regulate the primary effect of speech as such. City of Portland v. Tidyman , 306 Or. 174, 182-83, 759 P.2d 242 (1988) ; see Robertson , 293 Or. at 415, 649 P.2d 569 (distinguishing laws written in terms directed against speech as such from laws written in terms directed against a forbidden effect but including speech as an element). Petitioners argue that it is unclear whether IP 28 would permit some or all of those regulations, and the unqualified use of the word "regulate" in the caption does not nothing to signal that uncertainty to the voters.
The Attorney General responds that the common and ordinary definition of "regulate" is sufficient to inform voters of the potential reach of the measure.2 If we looked only at the word "regulate" in isolation, we might be inclined to agree with the Attorney General. However, our cases have explained that the context in which a word is used bears on its *971meaning. See State v. Cloutier , 351 Or. 68, 96, 261 P.3d 1234 (2011) (explaining that dictionaries "do not tell us what words mean, only what words can mean, depending on their context and the particular manner in which they are used") (emphasis in original). We agree with petitioners that the word "regulate," when used in the context of regulating expressive activity, can encompass a range of different types of regulations. Because the measure does not identify the extent to which IP 28 would permit **538the regulation of campaign contributions and expenditures, we think that the caption should signal that "regulate" is undefined and could encompass some or all of a range of regulatory measures. See Wolf v. Myers , 343 Or. 494, 501, 173 P.3d 812 (2007) (discussing ways in which ballot titles can signal that the terms used in a measure are either undefined or ambiguous).
Relying on Meyer v. Bradbury , 341 Or. 288, 297-98, 142 P.3d 1031 (2006), petitioner Lutz argues that the requirement that laws regulating campaign contributions and expenditures be passed by initiative or by an elected legislative body by a three-fourths vote is a major effect of IP 28 that should be included in the caption. As explained below, we conclude that that requirement is a significant effect of the measure that should be included in the "yes" result statement. However, we do not think that it is the "subject" or "major effect" of the measure. Rather, it is a subsidiary condition that must be met before a law may regulate campaign expenditures or contributions. In that regard, we note that our decision in Meyer , on which Lutz relies, involved a separate-amendment challenge to a measure, not a challenge to the caption of a ballot title. We have considered petitioners' other challenges to the caption and conclude either that the foregoing discussion obviates the need to address those challenges or that those challenges are not well-taken.
Petitioners also challenge the result statements. ORS 250.035(2)(b) and (c) provide that ballot titles must contain a "simple and understandable statement of not more than 25 words that describes the result if the measure is" approved or rejected. Generally, a ballot title's result statements must adequately describe the effects of voting one way or another. As this court has observed, the "yes" result statement should describe "the most significant and immediate" effects of the ballot initiative for "the general public." Novick/Crew v. Myers , 337 Or. 568, 574, 100 P.3d 1064 (2004). Similarly, a "no" result statement should "addres[s] the substance of current law on the subject matter of the proposed measure" and "summariz[e] the current law accurately." Id. at 577, 100 P.3d 1064 (emphasis omitted).
**539The Attorney General's certified "yes" result statement provides:
" 'Yes' vote amends constitution, allows laws regulating contributions and expenditures to influence election outcomes; triggers Measure 47 (from 2006), regulating campaign expenditures, requiring certain disclosures."
Petitioners challenge the "yes" result statement both for its statement that a "yes" vote would "trigge[r] Measure 47" and for its failure to signal that the extent to which IP 28 would permit "regulation" of campaign contributions and expenditures is unclear. We agree with those challenges for the reasons stated above.
Petitioner Lutz also argues that the "yes" result statement is deficient because it does not mention that laws regulating campaign expenditures and contributions must be passed "by initiative or by an elected legislative body by a three-fourths vote." Although the means by which those regulatory laws may be enacted is not the subject or major effect of the measure, we agree with Lutz that the three-fourths vote requirement is a significant effect of the measure. If IP 28 were adopted, it would restrict the customary ability of elected legislative bodies (and perhaps initiated measures) to enact regulatory laws by a majority vote. That result is sufficiently significant to require its mention in the "yes" result statement.
Petitioners Markley and Hedbor raise a related issue. They contend that the measure is ambiguous as to whether the three-fourths vote requirement applies only to laws passed by elected legislative bodies or whether that *972requirement also applies to laws adopted by initiative. As noted, IP 28 contains the following proviso: "provided, that such laws are adopted or amended by initiative or by an elected legislative body by a three-fourths vote." The Attorney General notes that the phrase "by a three-fourths vote" comes immediately after the phrase "by an elected legislative body." It follows, she argues, that the three-fourths vote requirement applies only to laws amended or adopted by elected legislative bodies. That is a reasonable reading of the proviso. **540However, the proviso lacks punctuation or other signals that would demonstrate that the three-fourths vote requirement would apply only to laws adopted by an elected legislative body. Cf. Meyer , 341 Or. at 292, 142 P.3d 1031 (setting out IP 8 (2006), which used commas to signal that the three-fourths vote requirement applied only to laws passed by the Legislative Assembly). Grammatically, the three-fourths vote requirement also could apply to laws adopted by initiative. Even if the Attorney General's reading were the better one, the proviso in IP 28 would still be ambiguous as to whether the three-fourths vote requirement applies only to laws adopted by elected legislative bodies or also to laws adopted by initiative. See Wolf , 343 Or. at 502, 173 P.3d 812 (finding an ambiguity in similar circumstances). As the court noted in Wolf , "it is sometimes possible to use the words of the measure itself * * * to describe the subject of an ambiguous measure while avoiding speculation about how the measure might be interpreted in some future legal proceeding." Id. at 503, 173 P.3d 812. Of course, the choice of how best to capture the ambiguity in the proviso remains with the Attorney General.
Petitioners Markley and Hedbor challenge the "no" result statement, which provides:
" 'No' vote retains Oregon Constitution's existing free-expression provision; Measure 47 (from 2006), regulating campaign expenditures, requiring certain disclosures, remains inoperative."
Markley and Hedbor argue that "the no result statement is tautological: saying nothing more than rejection of the constitutional amendment will cause the constitution not to be amended." Again, we agree. As the court explained in Novick/Crew , a "no" result statement should "addres[s] the substance of current law on the subject matter of the proposed measure." 337 Or. at 577, 100 P.3d 1064 (emphasis omitted). The result statement does not say what the substance of the current law is. It may be that stating that a "no" vote would leave Measure 47 (2006) inoperative illustrates a result of a "no" vote. However, in the same way that enacting IP 28 would authorize a class of regulatory laws rather than merely revive a single dormant law, voting "no" on IP 28 would have a broader effect than merely leaving a single law **541inoperative.3 A description of the substance of the current law is required.
We turn finally to petitioners' challenges to the summary. ORS 250.035(2)(d) requires that the ballot title contain a "concise and impartial statement of not more than 125 words summarizing the state measure and its major effect." The purpose of a ballot title's summary is to give voters enough information to understand what will happen if the initiative is adopted. Whitsett v. Kroger , 348 Or. 243, 252, 230 P.3d 545 (2010). For convenience, we again set out the summary for IP 28:
"Amends Constitution. The Oregon Supreme Court has interpreted the Oregon Constitution's free-expression provision ( Article I, section 8 ) to prohibit limits on many political campaign contributions and expenditures. The proposed measure amends Article I, section 8 to allow laws that regulate contributions/expenditures to influence elections. Such laws must be adopted or amended by initiative or by three-fourths vote by legislative body. If amendment passes, Measure 47 (enacted by voters in 2006) will become law and will *973alter Oregon's current election laws. Measure 47 limits campaign contributions/expenditures, establishes new reporting/advertising disclosure requirements for the sources and amounts of campaign contributions/expenditures. Laws regulating campaign contributions and expenditures (including Measure 47) still must comply with the First Amendment to the United States Constitution."
Petitioners challenge the summary for reasons that we have already discussed. They contend that the summary impermissibly resolves an ambiguity in the proviso to IP 28-whether the three-fourths vote requirement would apply to laws adopted by initiative as well as by elected legislative bodies. They also argue that the word "regulate" is insufficient to signal that the extent to which IP 28 would permit regulation of campaign contributions and expenditures is unclear. For the reasons stated above, we agree with those challenges.
**542Petitioners also reiterate their challenges to saying that, if IP 28 is adopted, Measure 47 (2006) will become law. Unlike the caption and the "yes" result statement, where we concluded that the effect of IP 28 on a specific law was either not the subject or a significant effect of the measure, we cannot say that the effect of IP 28 on Measure 47 (2006) is an inappropriate subject to discuss in the summary. That leaves us with the question whether the Attorney General is correct in stating, as she does in the summary, that, "[i]f [the] amendment passes, Measure 47 (enacted by the voters in 2006) will become law and will alter Oregon's current election laws."
Petitioners Markley and Hedbor argue that the Attorney General's prediction that Measure 47 (2006) will become law is "impermissible 'speculation about potential secondary effects' " of IP 28. (Quoting Rooney v. Kulongoski , 322 Or. 15, 37, 902 P.2d 1143 (1995).) They reason that, if IP 28 were adopted, it would apply prospectively only and would not retroactively revive past laws, such as Measure 47 (2006). Alternatively, they contend that permitting laws that "regulate" campaign contributions and expenditures (as IP 28 would) does not necessarily mean that those laws would also "limit" the amount of campaign contributions and expenditures (as Measure 47 (2006) does). In their view, regulating contributions and expenditures is not the same thing as limiting the amount contributed or expended.
Petitioner Lutz similarly argues that, because the effect of IP 28 on Measure 47 (2006) is, at best, uncertain, the Attorney General should neither speculate how IP 28 would affect Measure 47 (2006), nor should she say definitively that Measure 47 (2006) will go into effect. Lutz observes that, even if Measure 47's limitations on campaign contributions and expenditures would have complied with the First Amendment when the measure was put before the voters in 2006, it is extremely unlikely that those limits would survive First Amendment scrutiny now in light of the United States Supreme Court's intervening decisions in McCutcheon v. Federal Election Commission , --- U.S. ----, 134 S.Ct. 1434, 188 L. Ed. 2d 468 (2014), and Citizens United v. Federal Election Commission , 558 U.S. 310, 130 S.Ct. 876, 175 L. Ed. 2d 753 (2010). Given those federal decisions, Lutz **543argues that the Attorney General's statement that Measure 47 (2006) will become law is inaccurate and misleading.
The Attorney General responds that, "by providing constitutional authorization for laws that 'regulate contributions and expenditures ... to influence the outcome of any election,' [IP 28] effectively abrogates Vannatta 's holding that such limits violate the Oregon Constitution." (Ellipsis in original.) It follows, she reasons, that IP 28 will "provid[e] the constitutional authorization necessary for Measure 47 to automatically take effect, by its own terms." And she concludes that, even if Measure 47 (2006) may enjoy only a fleeting existence in light of the United States Supreme Court's decisions in Citizens United and McCutcheon , it would be inappropriate to pass on the constitutionality of Measure 47 in advance of a First Amendment challenge to it.
One difficulty with the Attorney General's reasoning lies in her conclusion that IP 28 will effectively abrogate Vannatta 's holding that contribution and expenditure limits violate Article I, section 8. The Attorney General's *974conclusion fails to account for the subject of the sentence that IP 28 would add to Article I, section 8. As noted, IP 28 provides:
"Laws consistent with the freedom of speech guarantee of the United States Constitution may regulate contributions and expenditures, of any type or description, to influence the outcome of any election; provided, that such laws are adopted or amended by initiative or by an elected legislative body by a three-fourths vote."
IP 28 does not provide that any law passed by the requisite three-fourths vote "may regulate [campaign] contributions and expenditures" under Article I, section 8. It provides that only those laws that are "consistent with the freedom of speech guarantee of the United States Constitution may regulate [campaign] contributions and expenditures." By its terms, IP 28 limits the laws that may regulate campaign contributions and expenditures under Article I, section 8, to "[l]aws consistent with [the First Amendment]." Or. at least that is a permissible reading of IP 28.
Under that reading, if Measure 47 (2006) is not consistent with the First Amendment (and the Attorney General does not dispute that there is a significant question whether **544it is), then it is not the sort of law that "may regulate campaign contributions and expenditures" under the terms of IP 28.4 Put differently, IP 28 may not abrogate very much, if any, of the holding in Vannatta I . Moreover, under the terms of Measure 47 (2006), that measure will be revived only if Article I, section 8, is amended to allow "limitations on political campaign contributions and expenditures," which this court explained referred to the "same kind of limitations struck down in Vannatta I. " Hazell , 352 Or. at 466, 287 P.3d 1079. We cannot say with any certainty that IP 28 would achieve that goal.
We accordingly agree with petitioners that, to the extent the summary mentions Measure 47 (2006), the most that the summary reasonably can say is that, if IP 28 were adopted, Measure 47 (2006) might be revived. For the reasons stated above, we refer to the caption, "yes" and "no" result statements, and the summary to the Attorney General for modification.
The certified ballot title is referred to the Attorney General for modification.
The court explained that the reference to " '[t]his Act' [in section 9(f) of Measure 47 (2006) ] [was] evidence that the voters intended and understood that, if and when Measure 47 became operative, it would do so as a whole piece of legislation, not in some piecemeal fashion." Hazell , 352 Or. at 465, 287 P.3d 1079.
The Attorney General also argues that petitioners Markley and Hedbor failed to raise that objection to the draft ballot title. We disagree. Before the Secretary of State, Markley and Hedbor objected to the "breadth" of the regulatory power that IP would permit, untethered to "the service of any specific purpose." We think their objections to the draft ballot title were sufficient to put the Attorney General on notice of their concerns regarding the undefined breadth of the term "regulate."
Although petitioners do not object to the reference to Measure 47 (2006) in the "no" result statement, we note that the implication of saying that a "no" vote will leave Measure 47 (2006) inoperative is that a "yes" vote would have the opposite result, a proposition that is not necessarily true.
We recognize that the terms of IP 28 invert the usual method of analysis in which we determine whether a statute violates the state constitution before determining whether it violates the federal constitution. However, IP 28 appears to be written in such a way that only those laws that are consistent with the First Amendment will come within the exception set out in IP 28 and thus avoid the limitations stated in Article I, section 8, as interpreted in Vannatta I .