delivered the opinion of the Court.
Since its enactment during the Civil War, the False Claims Act, 31 U. S. C. §§ 3729-3733, has authorized both the Attorney General and private qui tam relators to recover from persons who make false or fraudulent claims for payment to the United States. The Act now contains a provision barring qui tam actions based upon the public disclosure of allegations or transactions in certain specified sources. § 3730(e)(4)(A). The question before us is whether the reference to “administrative” reports, audits, and investigations in that provision encompasses disclosures made in state and local sources as well as federal sources. We hold that it does.1
I
In 1995 the United States Department of Agriculture (USDA) entered into contracts with two counties in North Carolina authorizing them to perform, or to hire others to perform, cleanup and repair work in areas that had suffered extensive flooding. The Federal Government agreed to shoulder 75 percent of the contract costs. Respondent Karen T. Wilson was at that time an employee of the Graham *284County Soil and Conservation District, a special-purpose government body that had been delegated partial responsibility for coordinating and performing the remediation effort. Suspecting possible fraud in connection with this effort, Wilson voiced her concerns to local officials in the summer of 1995. She also sent a letter to, and had a meeting with, agents of the USD A.
Graham County officials began an investigation. An accounting firm hired by the county performed an audit and, in 1996, issued a report (Audit Report) that identified several potential irregularities in the county’s administration of the contracts. Shortly thereafter, the North Carolina Department of Environment, Health, and Natural Resources issued a report (DEHNR Report) identifying similar problems. The USDA’s Office of Inspector General eventually issued a third report that contained additional findings.
In 2001 Wilson filed this action, alleging that petitioners, the Graham County and Cherokee County Soil and Water Conservation Districts and a number of local and federal officials, violated the False Claims Act (FCA) by knowingly submitting false claims for payment pursuant to the 1995 contracts. She further alleged that petitioners retaliated against her for aiding the federal investigation of those false claims. Following this Court's review of the statute of limitations applicable to Wilson’s retaliation claim, Graham County Soil & Water Conservation Dist. v. United States ex rel. Wilson, 545 U. S. 409 (2005), the Court of Appeals ordered that that claim be dismissed as time barred. 424 F. 3d 437 (CA4 2005). On remand, the District Court subsequently dismissed Wilson’s qui tarn action for lack of jurisdiction. App. to Pet. for Cert. 95a-105a. The court found that Wilson had failed to refute that her action was based upon allegations publicly disclosed in the Audit Report and the DEHNR Report. Id., at 95a-98a. Those reports, the District Court determined, constituted “administrative . . . report[s], . . . audit[s], or investigation[s]” within the *285meaning of the FCA’s public disclosure bar, 31 U. S. C. § 3730(e)(4)(A).
The Court of Appeals reversed the judgment of the District Court because the reports had been generated by state and local entities. “[O]nly federal administrative reports, audits or investigations,” the Fourth Circuit concluded, “qualify as public disclosures under the FCA.” 528- F. 3d 292, 301 (2008) (emphasis added). The Circuits having divided over this issue,2 we granted certiorari to resolve the conflict. 557 U. S. 918 (2009).
II
We have examined the FCA’s qui tam provisions in several recent opinions.3 At issue in this case is the FCA’s public disclosure bar, which deprives courts of jurisdiction over qui tam suits when the relevant information has already entered the public domain through certain channels. The statute contains three categories of jurisdiction-stripping disclo*286sures. Following the example of the Court of Appeals, see 528 F. 3d, at 300-301, we have inserted Arabic numerals to identify these categories:
“No court shall have jurisdiction over an action under this section based upon the public disclosure of allegations or transactions [1] in a criminal, civil, or administrative hearing, [2] in a congressional, administrative, or Government Accounting Office [(GAO)] report, hearing, audit, or investigation, or [3] from the news media, unless the action is brought by the Attorney General or the person bringing the action is an original source4 of the information.” § 3730(e)(4)(A) (footnote omitted).
This dispute turns on the meaning of the adjective “administrative” in the second category (Category 2): whether it embraces only forums that are federal in nature, as respondent alleges, or whether it extends to disclosures made in state and local sources such as the DEHNR Report and the Audit Report, as petitioners allege.
In debating this question, petitioners have relied primarily on the statute's text whereas respondent and the Solicitor General, as her amicus, have relied heavily on considerations of history and policy. Although there is some overlap among the three types of argument, it is useful to discuss them separately. We begin with the text.
III
The term “administrative” “may, in various contexts, bear a range of related meanings,” Chandler v. Judicial Council of Tenth Circuit, 398 U. S. 74, 103, n. 8 (1970) (Harlan, J., concurring in denial of writ), pertaining to private bodies as *287well as to governmental bodies. When used to modify the nouns “report, hearing, audit, or investigation,” in the context of a statutory provision about “the public disclosure” of fraud on the United States, the term is most naturally read to describe the activities of governmental agencies. See Black’s Law Dictionary 49 (9th ed. 2009) (hereinafter Black’s) (defining “administration,” “[i]n public law, [as] the practical management and direction of the executive department and its agencies”). Given that “administrative” is not itself modified by “federal,” there is no immediately apparent textual basis for excluding the activities of state and local agencies (or their contractors) from its ambit. As the Court of Appeals recognized, “the statute by its express terms does not limit its reach to federal administrative reports or investigations.” 528 F. 3d, at 301. “[T]here is nothing inherently federal about the word ‘administrative,’ and Congress did not define the term in the FCA.” Id., at 302.
The Court of Appeals’ conclusion that “administrative” nevertheless reaches only federal sources rested on its application of the interpretive maxim noscitur a sociis. See id., at 302-305. This maxim, literally translated as “ ‘it is known by its associates,’” Black’s 1160, counsels lawyers reading statutes that “a word may be known by the company it keeps,” Russell Motor Car Co. v. United States, 261 U. S. 514, 519 (1923). All participants in this litigation acknowledge that the terms “congressional” and “[GAO]” are federal in nature; Congress is the Legislative Branch of the Federal Government,5 and the GAO is a federal agency.6 Relying on *288our opinions in S. D. Warren Co. v. Maine Bd. of Environmental Protection, 547 U. S. 370 (2006), and Beecham v. United States, 511 U. S. 368 (1994), the Court of Appeals reasoned that “the placement of ‘administrative’ squarely in the middle of a list of obviously federal sources strongly suggests that ‘administrative’ should likewise be restricted to federal administrative reports, hearings, audits, or investigations.” 528 F. 3d, at 302. In so holding, the Court of Appeals embraced what we might call the Sandwich Theory of the Third Circuit. Both courts ‘“f[ou]nd it hard to believe that the drafters of this provision intended the word “administrative” to refer to both state and federal reports when it lies sandwiched between modifiers which are unquestionably federal in character.’” Ibid, (quoting United States ex rel. Dunleavy v. County of Delaware, 123 F. 3d 734, 745 (CA3 1997)).
We find this use of noscitur a sociis unpersuasive. A list of three items, each quite distinct from the other no matter how construed, is too short to be particularly illuminating. Although this list may not be “completely disjunctive,” 528 F. 3d, at 302 — it refers to “congressional, administrative, or [GAO]” sources, § 3730(e)(4)(A), rather than “congressional, or administrative, or [GAO]” sources — neither is it completely harmonious. The substantive connection, or fit, between the terms “congressional,” “administrative,” and “GAO” is not so tight or so self-evident as. to demand that we “rob” any one of them “of its independent and ordinary significance.” Reiter v. Sonotone Corp., 442 U. S. 330, 338-339 (1979); see also Russell, 261 U. S., at 519 (“That a word may be known by the company it keeps is .. . not an invariable rule, for the word may have a character of its own not to be submerged by its association”). The adjectives in Cat*289egory 2 are too few and too disparate to qualify as “a string of statutory terms,” S. D. Warren Co., 547 U. S., at 378, or “items in a list,” Beecham, 511 U. S., at 371, in the sense that we used those phrases in the cited cases.7
More importantly, we need to evaluate “administrative” within the larger scheme of the public disclosure bar. Both parties acknowledge, as they must, that “Statutory language has meaning only in context,” Graham County Soil, 545 U. S., at 415; where they differ is in determining the relevant context. The Sandwieh Theory presupposes that Category 2 is the only piece of § 3730(e)(4)(A) that matters. We agree with petitioners, however, that all of the sources listed in § 3730(e)(4)(A) provide interpretive guidance. All of these sources drive at the same end: specifying the types of disclosures that can foreclose qui tarn actions. In light of the public disclosure bar's grammatical structure, it may be convenient and even clarifying to parse the list of sources into three categories. But it does not follow that we should *290treat these categories as islands unto themselves. Courts have a “duty to construe statutes, not isolated provisions.” Gustafson v. Alloyd Co., 513 U. S. 561, 568 (1995).
When we consider the entire text of the public disclosure bar, the case for limiting “administrative” to federal sources becomes significantly weaker. The “news media” referenced in Category 3 plainly have a broader sweep. The Federal Government fluids certain media outlets, and certain private outlets have a national focus; but no one contends that Category 3 is limited to these sources. There is likewise no textual basis for assuming that the “criminal, civil, or administrative hearing[s]” listed in Category 1 must be federal hearings.8 Of the numerous types of sources that serve a common function in § 3730(e)(4)(A), then, only two are distinctly federal in nature, while one (the news media) is distinctly nonfederal in nature.
If the Court of Appeals was correct that the term “administrative” encompasses state and local sources in Category 1, see 528 F. 3d, at 303, it becomes even harder to see why the term would not do the same in Category 2. See Erlenbaugh v. United States, 409 U. S. 239, 243 (1972) (“[A] legislative body generally uses a particular word with a consistent meaning in a given context”). Respondent and the Solicitor General assert that § 3730(e)(4)(A)’s two references to “administrative” can be distinguished because Category 1 is best understood to refer to adjudicative proceedings, whereas Category 2 is best understood to refer to legislative or quasi-legislative activities such as rulemaking, oversight, and investigations. See Brief for Respondent 16-18; Brief *291for United States as Amicus Curiae 25-26 (hereinafter Brief for United States). Yet even if this reading were correct, state and local administrative reports, hearings, audits, and investigations of a legislative-type character are presumably just as public, and just as likely to put the Federal Government on notice of a potential fraud, as state and local administrative hearings of an adjudicatory character.9
Respondent and the Solicitor General try to avoid this inference, and to turn a weakness into a strength, by further averring that the sources listed in Category 1 are themselves only federal. See Brief for Respondent 23-24; Brief for United States 25-26. No court has ever taken such a view of these sources. See 528 F. 3d, at 303 (citing cases from the Third, Fourth, Fifth, Ninth, and Eleventh Circuits and stating that “[t]he courts have easily concluded that [Category 1] applies to state-level hearings”); Sylvia § 11:37, at 643, n. 1 (citing additional cases).10 The arguments in favor of read*292ing a federal limitation into Category 1 are supported, if at all, by legislative history and policy; they find no support in the statute’s text.
Moving from the narrow lens of the Sandwich Theory to a bird’s-eye view, respondent and the Solicitor General also maintain that the “exclusively federal focus” of the FCA counsels against reading the public disclosure bar to encompass nonfederal sources. Brief for Respondent 10,18; Brief for United States 13. The FCA undoubtedly has a federal focus. But so does every other federal statute. And as respondent and the Solicitor General elsewhere acknowledge, quite a few aspects of the FCA, including a reference to “administrative” proceedings in § 3733(i)(7)(A)11 and the reference to “news media” in § 3730(e)(4)(A) itself, are not just federal. In any event, the “federal focus” of the statute, as a whole, does not shine light on the specific question whether the public disclosure bar extends to certain nonfederal contexts. It is the fact of “public disclosure” — not Federal Government creation or receipt — that is the touchstone of § 3730(e)(4)(A).
Respondent and the Solicitor General make one last argument grounded in the statutory text: It would be anomalous, they say, for state and local administrative reports to count as public disclosures, when state legislative reports do not. See Brief for Respondent 15; Brief for United States 15-16. Yet neither respondent nor the Solicitor General disputes the contention of petitioners and their amici that, at the time the public disclosure bar was enacted in 1986, Congress *293rarely gave state legislatures a meaningful role in administering or overseeing federally funded programs. See Brief for Petitioners 36-39; Brief for National League of Cities et al. as Amici Curiae 8-13. As in the instant ease, the Federal Government was far more likely to enter into contracts with, and to provide moneys to, state and local executive agencies. Whether or not state legislative sources should have been included in § 3730(e)(4)(A), their exclusion therefore carries no clear implications for the status of state administrative sources.
In sum, although the term “administrative” may be sandwiched in Category 2 between terms that are federal in nature, those terms are themselves sandwiched between phrases that have been generally understood to include non-federal sources; and one of those phrases, in Category 1, contains the exact term that is the subject of our inquiry. These textual clues negate the force of the noscitur a sociis canon, as it was applied by the Court of Appeals.12 We are not persuaded that the associates with which “administrative” keeps company in § 3730(e)(4)(A) endow it with an exclusively federal character.
IV
As originally enacted, the FCA did not limit the sources from which a relator could acquire the information to bring *294a qui tam action. In United States ex rel. Marcus v. Hess, 317 U. S. 537 (1943), we upheld the relator’s recovery even though he had discovered the fraud by reading a federal criminal indictment — a quintessential “parasitic” suit. Id., at 545-548; see id., at 545 (“Even if, as the government suggests, the petitioner has contributed nothing to the discovery of this crime, he has contributed much to accomplishing one of the purposes for which the Act was passed”). Congress promptly reacted to that decision by amending the statute to preclude qui tam actions “based upon evidence or information in the possession of the United States, or any agency, officer or employee thereof, at the time such suit was brought.” Act of Dec. 23, 1943, 57 Stat. 609 (codified at 31 U. S. C. § 232(C) (1946 ed.)). This amendment erected what came to be known as a Government knowledge bar: “[0]nce the United States learned of a false claim, only the Government could assert its rights under the FCA against the false claimant.” Hughes Aircraft Co. v. United States ex rel. Schumer, 520 U. S. 939, 949 (1997) (internal quotation marks omitted). In the years that followed the 1943 amendment, the volume and efficacy of qui tam litigation dwindled. “Seeking the golden mean between adequate incentives for whistle-blowing insiders with genuinely valuable information and discouragement of opportunistic plaintiffs who have no significant information to contribute of their own,” United States ex rel. Springfield Terminal R. Co. v. Quinn, 14 F. 3d 645, 649 (CADC 1994), Congress overhauled the statute once again in 1986 “to make the FCA a ‘more useful tool against fraud in modern times,’ ” Cook County v. United States ex rel. Chandler, 538 U. S. 119, 133 (2003) (quoting S. Rep. No. 99-345, p. 2 (1986) (hereinafter S. Rep.)).
The present text of § 3730(e)(4) was enacted in 1986 as part of this larger reform. Congress apparently concluded that a total bar on qui tam actions based on information already in the Government’s possession thwarted a significant number of potentially valuable claims. Rather than simply repeal the Government knowledge bar, however, Congress re*295placed it with the public disclosure bar in an effort to strike a balance between encouraging private persons to root out fraud and stifling parasitic lawsuits such as the one in Hess. How exactly § 3730(e)(4) came to strike this balance in the way it did is a matter of considerable uncertainty. The House and Senate Judiciary Committees each reported bills that contained very different public disclosure bars from the one that emerged in the Statutes at Large; the Senate bill, for example, did not include the words “administrative,” “audit,” or “investigation” in its version of Category 2, nor did it contain an original source exception. See S. Rep., at 42-43 (text of proposed § 3730(e)(4)).13
In respondent and her amici’s view, this background counsels in favor of an exclusively federal interpretation of “administrative” for three separate reasons. First, the drafting history of the public disclosure bar suggests that Congress intended such a result. Second, a major aim of the 1986 amendments was to limit the scope of the Government knowledge bar, and “[cjonstruing [§ 3730(e)(4)(A)] as limited to disclosures in federal proceedings furthers Congress’s purpose 'to encourage more private enforcement suits.’ ” Brief for United States 21 (quoting S. Rep., at 23-24). Third, whereas federal administrative proceedings can be presumed to provide the Attorney General with a fair opportunity to decide whether to bring an FCA action based on revelations made therein, the Attorney General is much less likely to learn of fraud disclosed in state proceedings. Respondent and her amici further maintain that it would be perverse to include nonfederal sources in Category 2, as local governments would then be able to shield themselves from qui tarn liability by discretely disclosing evidence of fraud in “public” reports.14
*296These arguments are reasonable so far as they go, but they do not go very far. As many have observed, the drafting history of the public disclosure bar raises more questions than it answers.15 Significant substantive changes — including the introduction of the term we are construing in this case — were inserted without floor debate, as “technical” amendments. That the original Senate bill mentioned only congressional and GAO sources in Category 2 is therefore of little moment. Neither the House nor the Senate Committee Report explained why a federal limitation would be appropriate, and the subsequent addition of “administrative” sources to this Category might be taken as a sign that such a limitation was rejected by the full Chambers.16
*297Respondent and her amici place particular emphasis on a remark made by the lead sponsor of the Senate bill, Senator Grassley. See Brief for Respondent 29; Brief for United States 20; Brief for American Center for Law and Justice as Amicus Curiae 13-14; Brief for Taxpayers Against Fraud Education Fund as Amicus Curiae 30-31. In a floor statement, Grassley said that “the term ‘Government’ in the definition of original source is meant to include any Government source of disclosures cited in [the public disclosure bar]; that is, Government includes Congress, the General Accounting Office, any executive or independent agency as well as all other governmental bodies that may have publicly disclosed the allegations.” 132 Cong. Rec. 20536 (1986). Yet even if a single sentence by a single legislator were entitled to any meaningful weight, Senator Grassley’s remark merely begs the question before us. His formulation fails to indicate whether the “other governmental bodies” may be state or local bodies. It also turns on a term, “Government” with a capital “G,” that does not appear in the codified version of the public disclosure bar, which Congress subsequently revised in numerous respects prior to passage.
There is, in fact, only one item in the legislative record that squarely corroborates respondent’s reading of the statute: a letter sent by the primary sponsors of the 1986 amendments to the Attorney General in 1999. See 145 Cong. Rec. 16032 (1999) (reproducing text of letter in which Rep. Berman and Sen. Grassley state: “We did intend, and any fair reading of the statute will confirm, that the disclosure must *298be in a federal criminal, civil or administrative hearing. Disclosure in a state proceeding of any kind should not be a bar to a subsequent qui tam suit”). Needless to say, this letter does not qualify as legislative “history,” given that it was written 13 years after the amendments were enacted. It is consequently of scant or no value for our purposes.17
We do not doubt that Congress passed the 1986 amendments to the FCA “to strengthen the Government’s hand in fighting false claims,” Cook County, 538 U. S., at 133-134, and “to encourage more private enforcement suits,” S. Rep., at 23-24. It is equally beyond cavil, however, that Congress passed the public disclosure bar to bar a subset of those suits that it deemed unmeritorious or downright harmful. The question before us concerns the precise scope of that subset; and on this matter, the record is all but opaque. While “the absence of specific legislative history in no way modifies the conventional judicial duty to give faithful meaning to the language Congress adopted in the light of the evident legislative purpose in enacting the law in question,” United States v. Bornstein, 423 U. S. 303, 310 (1976), there is no “evi*299dent legislative purpose” to guide our resolution of the discrete issue that confronts us.
V
Respondent and her amici likewise fail to prove their case that petitioners’ reading of the statute will lead to results that Congress could not have intended. Their argument rests on an empirical proposition: “While federal inquiries and their outcomes are readily available to Department of Justice [(DOJ)] attorneys, many state and local reports and investigations never come to the attention of federal authorities.” Brief for United States 22; see also 528 F. 3d, at 306 (“Because the federal government is unlikely to learn about state and local investigations, a large number of fraudulent claims against the government would go unremedied without the financial incentives offered by the qui tarn provisions of the FCA”). This proposition is not implausible, but it is sheer conjecture. Numerous federal investigations may be occurring at any given time, and DOJ attorneys may not reliably learn about their findings. DOJ attorneys may learn about quite a few state and local inquiries, especially when the inquiries are conducted pursuant to a joint federal-state program financed in part by federal dollars, such as the program at issue in this case.18 Just how accessible to the Attorney General a typical state or local source will be, as compared to a federal source, is an open question. And it is *300not even the right question. The statutory touchstone, once again, is whether the allegations of fraud have been “public[ly] disclosed],” § 3730(e)(4)(A), not whether they have landed on the desk of a DOJ lawyer.
Respondent’s argument also gives insufficient weight to Congress’ decision to bar qui tam actions based on disclosures “from the news media.” Ibid. Because there was no such bar prior to 1986, the addition of the news media as a jurisdiction-stripping category forecloses the suggestion that the 1986 amendments implemented a single-minded intent to increase the availability of qui tam litigation. And since the “news media” include a large number of local newspapers and radio stations, this category likely describes a multitude of sources that would seldom come to the attention of the Attorney General.
As for respondent and her amici’s concern that local governments will insulate themselves from qui tam liability “through careful, low key ‘disclosures’” of potential fraud, Brief for American Center for Law and Justice as Amicus Curiae 17, this argument rests not just on speculation but indeed on rather strained speculation. Any such disclosure would not immunize the local government from FCA liability in an action brought by the United States, see Rockwell Int’l Corp. v. United States, 549 U. S. 457, 478 (2007) — and to the contrary it could tip off the Attorney General that such an action might be fruitful. It seems to us that petitioners have the more clear-eyed view when they assert that, “[g]iven the fact that the submission of a false claim to the United States subjects a defendant to criminal liability, fines, debarment, treble damages and attorneys’ fees, no rational entity would prepare a report that self-discloses fraud with the sole purpose of cutting off qui tam actions.” Reply Brief for Petitioners 19; see also United States ex rel. Bly-Magee v. Premo, 470 F. 3d 914, 919 (CA9 2006) (“The fear [of self-insulating disclosures] is unfounded in general because it is unlikely that an agency trying to cover up its fraud would *301reveal the requisite 'allegations or transactions’ underlying the fraud in a public document”).19
Our conclusion is buttressed by the fact that Congress carefully preserved the rights of the most deserving qui tam plaintiffs: those whistle-blowers who qualify as original sources. Notwithstanding public disclosure of the allegations made by a qui tam plaintiff, her case may go forward if she is “an original source of the information.” § 3730(e)(4)(A). It is therefore flat wrong to suggest that a finding for petitioners will “ 'in effect return us to the unduly restrictive “government knowledge” standard’” that prevailed prior to 1986. Brief for United States 31 (quoting Dunleavy, 123 F. 3d, at 746); see Brief for Respondent 34 (asserting that “petitioners’ construction would reimpose a form of the 'government knowledge’ bar” (capitalization omitted)). Today’s ruling merely confirms that disclosures made in one type of context — a state or local report, audit, or investigation — may trigger the public disclosure bar. It has no bearing on disclosures made in other contexts, and it leaves intact the ability of original sources to prosecute qui tam actions irrespective of the state of Government knowledge. Whether respondent can qualify as an “original source,” as that term is defined in § 3730(e)(4), is one of many issues that remain open on remand.
VI
Respondent and the Solicitor General have given numerous reasons why they believe their reading of the FCA *302moves it closer to the golden mean between an inadequate and an excessive scope for private enforcement. Congress may well have endorsed those views in its recent amendment to the public disclosure bar. See n. 1, supra. With respect to the version of § 3730(e)(4)(A) that is before us, however, we conclude that the term “administrative” in Category 2 is not limited to federal sources.
The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.
It is so ordered.
On March 23, 2010, the President signed into law the Patient Protection and Affordable Care Act, Pub. L. 111-148, 124 Stat. 119. Section 10104(j)(2) of this legislation replaces the prior version of 31 U. S. C. 13730(e)(4) with new language. The legislation makes no mention of retroactivity, which would be necessary for its application to pending cases given that it eliminates petitioners’ claimed defense to a qui tam suit. See Hughes Aircraft Co. v. United States ex rel. Schumer, 520 U. S. 939, 948 (1997). Throughout this opinion, we use the present tense in discussing the statute as it existed at the time this case was argued.
Compare 528 P. 3d, at 301-307 (limiting this portion of the public disclosure bar to federal sources), and United States ex rel. Dunleavy v. County of Delaware, 123 P. 3d 734, 745-746 (CA3 1997) (same), with United States ex rel. Bly-Magee v. Premo, 470 P. 3d 914, 918-919 (CA9 2006) (concluding that state and local sources may qualify), cert. denied, 552 U. S. 1165 (2008), and Battle v. Board of Regents for State of Ga., 468 P. 3d 755, 762 (CA11 2006) (per curiam) (assuming without analysis that state audits may qualify). The Eighth Circuit appears to have taken a “middle road” on this issue, 528 F. 3d, at 301, holding that disclosures made in nonfederal forums may count as “ ‘administrative . . . report[s]’ ” or “ ‘audit[s]’ ” under § 3730(e)(4)(A) in some instances, as when they relate to “a cooperative federal-state program through which the federal government provides financial assistance,” Hays v. Hoffman, 325 F. 3d 982, 989, cert. denied, 540 U. S. 877 (2003).
See, e. g., Rockwell Int’l Corp. v. United States, 549 U. S. 457 (2007) (construing § 3730(e)(4)(A)’s original source exception); Cook County v. United States ex rel. Chandler, 538 U. S. 119 (2003) (holding that local governments are subject to qui tam liability); Vermont Agency of Natural Resources v. United States ex rel. Stevens, 529 U. S. 765 (2000) (holding that States are not subject to private FCA actions).
A separate statutory provision defines an “original source” as “an individual who has direct and independent knowledge of the information on which the allegations are based and has voluntarily provided the information to the Government before filing an action under this section which is based on the information.” 31 U. S. C. § 3730(e)(4)(B).
See U. S. Const., Art. I, §1; id., §4, cl. 1 (distinguishing “State . . . Legislature[sJ” from “the Congress”).
The statute refers to the GAO, mistakenly, as the “Government Accounting Office.” It is undisputed that the intended referent was the General Accounting Office, now renamed the Government Accountability Office. See 31 U. S. C. § 3730, p. 254, n. 2 (compiler’s note); 528 F. 3d 292, 300, n. 4 (CA4 2008); United States ex rel. Mistick PBT v. Housing Authority of Pittsburgh, 186 F. 3d 376, 387 (CA3 1999) (Alito, J.), cert. denied, *288529 U. S. 1018 (2000); see also Mistick, 186 F. 3d, at 398 (Becker, C. J., dissenting) (noting that courts have “frequently” made the same scrivener’s error). We have described the GAO as “an independent agency within the Legislative Branch that exists in large part to serve the needs of Congress.” Bowsher v. Merck & Co., 460 U. S. 824, 844 (1983).
In Jarecki v. G. D. Searle & Co., 367 U. S. 303 (1961), the Court applied the noscitur a sociis maxim in construing a statutory provision that referred to ‘“[i]ncome resulting from exploration, discovery, or prospecting,’” id., at 305 (quoting §456(a)(2)(B) of the Internal Revenue Code of 1939). Justice Sotomayor contends that “the three terms in Category 2 are no more ‘distinct’ or ‘disparate’ than the phrase at issue in Jarecki.” Post, at 305-306 (dissenting opinion) (citation omitted). We disagree. Whether taken in isolation or in context, the phrase “congressional, administrative, or GAO” is not as cohesive as the phrase “exploration, discovery, or prospecting. ” That is one reason why noscitur a sociis proved illuminating in Jarecki, and why it is less helpful in this case. On their “face,” the terms “exploration,” “discovery,” and “prospecting” all describe processes of searching, seeking, speculating; the centrality of such activities to “the oil and gas and mining industries” gave a due that it was those industries Congress had in mind when it drafted the provision. 367 U. S., at 307 (internal quotation marks omitted). The terms “congressional,” “administrative,” and “GAO” do not share any comparable core of meaning — or indeed any “common feature” at all, post, at 306 — apart from a governmental connotation. It takes the Sandwich Theory to graft a federal limitation onto “administrative.”
A number of lower courts have concluded that, as used in Category 1, “‘hearing’ is roughly synonymous with ‘proceeding.’” United, States ex rel. Springfield Terminal R. Co. v. Quinn, 14 E 3d 645, 652 (CADC 1994); see also 1 J. Boese, Civil False Claims and Qui Tam Actions §4.02[B], p. 4-59, and n. 231 (3d ed. 2006) (hereinafter Boese); C. Sylvia, The False Claims Act: Fraud Against the Government § 11:35, p. 642 (2004) (hereinafter Sylvia).
See Bly-Magee, 470 F. 3d, at 918 (“Indeed, the statute would seem to be inconsistent if it included state and local administrative hearings as sources of public disclosures [in Category 1] and then, in the next breath, excluded state administrative reports as sources"); In re Natural Gas Royalties Qui Tam Litigation, 467 F. Supp. 2d 1117, 1143-1144 (Wyo. 2006) (“There is no reason to conclude that Congress intended to limit administrative reports, audits, and investigations to federal actions, while simultaneously allowing all state and local civil litigation, state and local administrative hearings, and state and local news media to be treated as public disclosures. To interpret the statute so narrowly would have the anomalous result of allowing public disclosure status to the most obscure local news report and the most obscure state and local civil lawsuit or administrative hearing, but denying public disclosure status to a formal public report of a state government agency”).
Following the Court of Appeals, see 528 F. 3d, at 303, respondent asserts that only the Ninth Circuit, in A-l Ambulance Serv., Inc. v. California, 202 F. 3d 1238, 1244 (2000), has explicitly considered and rejected the argument that Category 1 is limited to federal sources. Brief for Respondent 23-24. At least one other Circuit, however, has done the same, see United States ex rel. Hafter v. Spectrum Emergency Care, Inc., 190 F. 3d 1156, 1161, n. 6 (CA10 1999), and no lower court, as far as we are aware, has so much as suggested that an alternative construction might be *292viable. Moreover, the Third, Fifth, and Eleventh Circuit cases cited by the Court of Appeals postdate A-l Ambulance and Dunleavy, 123 F. 3d 734, both of which put litigants and courts on notice of the possibility that § 3730(e)(4)(A) might be limited to federal sources.
On its face, § 3733(i)(7)(A) is silent as to whether it includes nonfederal proceedings. Respondent and the Solicitor General suggest that it does, though they fairly argue that this provision, relating to civil investigative demands, has little if any relevance to the case at hand. See Brief for Respondent 21, n. 8; Brief for United States 31-32.
The Court of Appeals repeatedly referred to the three categories in § 3730(e)(4)(A) as “clauses.” See 528 F. 3d, at 300-305. Were they in fact clauses rather than prepositional phrases, reliance on noscitur a sociis might have been supported by one of our earliest cases using that term, Watson v. Mercer, 8 Pet. 88,105 (1884) (Reporter’s statement of the ease), which suggested that “different clauses of the same sentence” should be presumed “to embrace the subject matter of the sentence.” The Court of Appeals’ mistaken reference to “clauses” is of course less significant than its failure to treat the public disclosure bar as an integrated whole. Cf. Stevens, The Shakespeare Canon of Statutory Construction, 140 U. Pa. L. Rev. 1373, 1376 (1992) (emphasizing importance of reading provisions in their broader statutory context).
See also H. R. Rep. No. 99-660, pp. 2-3 (1986) (text of proposed § 3730(b)(5)). The public disclosure bar that was enacted more closely resembles the version in the Senate bill.
State governments are already shielded from qui tam liability under our precedent. Stevens, 529 U. S. 765.
See, e. g., Dunleavy, 123 F. 3d, at 745 (“Congress gave us little specific guidance to determine the scope of public disclosure sources”); United States ex rel. Stinson, Lyons, Gerlin & Bustamante, P. A. v. Prudential Ins. Co., 944 F. 2d 1149, 1154 (CA3 1991) (“The bill that eventuated in the 1986 amendments underwent substantial revisions during its legislative path. This provides ample opportunity to search the legislative history and find some support somewhere for almost any construction of the many ambiguous terms in the final version”); id., at 1163 (Scirica, J., dissenting) (“One difficulty in interpreting the 1986 amendments is that Congress was never completely dear about what kind of ‘parasitic’ suits it was attempting to avoid”); Boese §4.02[A], at 4-46 (“The present Section 3730(e)(4) was enacted... without explanation by Congress”); id., § 4.02[A], at 4-47 to 4-48 (“[A]pplicable legislative history explaining versions [of § 3730(e)(4)] not adopted is of little help in dedphering this provision. Because Section 3730(e)(4) was drafted subsequent to the completion of the House and Senate Committee reports on the proposed False Claims Act Amendments, those reports, which contained discussion of altogether different bars, cannot be used in interpreting it. And the sponsors’ interpretations of the provision ultimately enacted ... are spare, often incorrect, and wide-ranging enough to provide some support for almost any construction of its many ambiguities”).
Justice Sotomayor makes a valiant effort to unearth from the legislative history “the balance Congress evidently sought to achieve through the 1986 amendments.” Post, at 312. But her reconstruction of the history assigns little weight to the side of this balance preserved by the public disclosure bar: the desire to minimize “the potential for parasitic *297lawsuits by those who learn of the fraud through public channels and seek remuneration although they contributed nothing to the exposure of the fraud,” United States ex rel. Doe v. John Doe Corp., 960 F. 2d 318, 319 (CA2 1992). And her narrative contains no account of why Category 2 emerged in the form that it did. Any such account would necessarily be an exercise in speculation, as the record is silent on the matter. In our view, neither the general trajectory of 20th-century FCA reform nor the specific statements made during the 1986 legislative process clearly point one way or the other on the question before us.
See Consumer Product Safety Comm’n v. GTE Sylvania, Inc., 447 U. S. 102, 118 (1980); Hamdan v. Rumsfeld, 548 U. S. 557, 580, n. 10 (2006); see also Hafter, 190 F. 3d, at 1161, n. 6 (refusing to credit the Berman-Grassley letter in interpreting the public disclosure bar). Respondent and her amici additionally contend that the enactment of the Program Fraud Civil Remedies Act of 1986 (PFCRA), 100 Stat. 1934 (codified at 31 U. S. C. § 3801 et seq.), shortly before the enactment of the FCA amendments supports their reading of the latter. See Brief for Respondent 30-33; Brief for United States 14-15; Brief for Taxpayers Against Fraud Education Fund as Amicus Curiae 28-29. Yet while “there is no question that the PFCRA was designed to operate in tandem with the FCA,” Stevens, 529 U. S., at 786, n. 17, or that the PFCRA is addressed to federal administrative agencies, there is also no explicit evidence to suggest that Congress intended to limit Category 2’s reference to “administrative” sources to the same set of agencies. The FCA's public disclosure bar serves a distinct function not replicated in the PFCRA; the text of the public disclosure bar contains no reference to the PFCRA; and no Member of Congress, so far as we are aware, articulated any such intent.
In some instances, federal law dictates that state and local governments receiving federal funds perform an audit of their programs. See 31 U. S. C. § 7502(a)(1)(B) (requiring nonfedera! entities that expend federal awards above a certain amount to “undergo a single audit” in accordance with specified conditions); Brief for State of Pennsylvania et al. as Amici Curiae 7-10 (discussing the Single Audit Act of 1984). It bears mention that, to the extent one is worried about Federal Government ignorance of state and local antifraud efforts, see post, at 312-313 (opinion of Sotomayoe, J.), today’s ruling may induce federal authorities to pay closer attention to such efforts going forward.
Petitioners and their amici also counter with public policy arguments of their own. Under the Court of Appeals’ reading of the statute, they allege, there is an increased likelihood that parasitic relators will beat more deserving relators to the courthouse, Brief for Petitioners 31, and that state and local governments will find their antifraud investigations impeded, or will decline to conduct such investigations in the first place, on account of “opportunistic potential relators trolling state records and reports, available to the public,” in search of a qui tarn claim, Brief for Commonwealth of Pennsylvania et al. as Amici Curiae 11.