with whom Justice Marshall joins, concurring in the judgment.
I agree with the Court that the construction given 18 U. S. C. § 1584 by the District Court and the Government either sweeps beyond the intent of Congress or fails to define the criminal conduct with sufficient specificity, and that a new trial under different instructions is therefore required. I cannot, however, square thé Court’s decision to add a physical or legal coercion limitation to the statute with either the statutory text or legislative history, and would adopt a different statutory construction that, I think, defines the crime with sufficient specificity but comports better with the evident intent of Congress.
*9541 — t
It is common ground among the parties and all the courts and Justices that have interpreted § 15841 that it encompasses, at a minimum, the compulsion of labor via the use or threat of physical or legal coercion. That much need not be belabored, for the use of the master’s whip and the power of the State to compel one human to labor for another were clearly core elements of slavery that the Thirteenth Amendment and its statutory progeny intended to eliminate. As the Government points out, however, the language of both the Thirteenth Amendment and § 1584 simply prohibits “involuntary servitude” and contains no words limiting the prohibition to servitude compelled by particular methods. “[The Thirteenth] amendment denounces a status or condition, irrespective of the manner or authority by which it is created.” Clyatt v. United States, 197 U. S. 207, 216 (1905).
*955If as a factual matter the use or threat of physical or legal coercion were the only methods by which a condition of involuntary servitude could be created, then the constitutional and statutory text might provide some support for the Court’s conclusion. But the Court does not dispute that other methods can coerce involuntary labor — indeed it is precisely the broad range of nonphysical private activities capable of coercing labor that the Court cites as the basis for its vagueness concerns. See ante, at 949; see also n. 5, infra. I address those concerns below, but the point here is only that those concerns, however serious, are not textual concerns, for the text suggests no grounds for distinguishing among different means of coercing involuntary servitude. Nor do I know of any empirical grounds for assuming that involuntary servitude can be coerced only by physical or legal means.2 To the contrary, it would seem that certain psychological, economic, and social means of coercion can be just as effective as physical or legal means, particularly where the victims are especially vulnerable, such as the mentally disabled victims in this case. Surely threats to burn down a person’s home or business or to rape or kill a person’s spouse or children can have greater coercive impact than the mere threat of a beating, yet the coercive impact of such threats turns not on any direct physical effect that would be felt by the laborer but on the psychological, emotional, social, or economic injury the *956laborer would suffer as a result of harm to his or her home, business, or loved ones. And drug addiction or the weakness resulting from a lack of food, sleep, or medical care can eliminate the will to resist as readily as the fear of a physical blow. Hypnosis, blackmail, fraud, deceit, and isolation are also illustrative methods — but it is unnecessary here to canvas the entire spectrum of nonphysical machinations by which humans coerce each other. It suffices to observe that one can imagine many situations in which nonphysical means of private coercion can subjugate the will of a servant.
Indeed, this case and others readily reveal that the typical techniques now used to hold persons in slavelike conditions are not limited to physical or legal means. The techniques in this case, for example, included disorienting the victims with frequent verbal abuse and complete authoritarian domination; inducing poor health by denying medical care and subjecting the victims to substandard food, clothing, and living conditions; working the victims from 3 a.m. to 8:30 p.m. with no days off, leaving them tired and without free time to seek alternative work; denying the victims any payment for their labor; and active efforts to isolate the victims from contact with outsiders who might help them.3 Without considering these techniques (and their particular effect on a mentally disabled person), one would hardly have a complete picture of whether the coercion inflicted on the victims was sufficient to *957make their servitude involuntary. Other involuntary servitude cases have also chronicled a variety of nonphysical and nonlegal means of coercion including triekery; isolation from friends, family, transportation or other sources of food, shelter, clothing, or jobs; denying pay or creating debt that is greater than the worker’s income by charging exorbitant rates for food, shelter, or clothing; disorienting the victims by placing them in an unfamiliar environment, barraging them with orders, and controlling every detail of their lives; and weakening the victims with drugs, alcohol, or by lack of food, sleep, or proper medical care. See, e. g., United States v. Warren, 772 F. 2d 827 (CA11 1985); United States v. Mussry, 726 F. 2d 1448 (CA9 1984); United States v. Ingalls, 73 F. Supp. 76 (SD Cal. 1947). One presumes these methods of coercion would not reappear with such depressing regularity if they were ineffective.4
My reading of the statutory language as not limited to physical or legal coercion is strongly bolstered by the legislative history. Section 1584 was created out of the consolidation of the Slave Trade statute and the Padrone statute. Ante, at 945. I agree with the Government that the background of both those statutes suggests that Congress intended to protect persons subjected to involuntary servitude by forms of coercion more subtle than force. The Padrone statute, for example, was designed to outlaw what was known as the “padrone system” whereby padrones in Italy inveigled from their parents young boys whom the padrones then used without pay as beggars, bootblacks, or street musicians. Once in this country, without relatives to turn to, the children had little choice but to submit to the demands of those asserting authority over them, yet this form of coercion was deemed sufficient — without any evidence of physical or legal coercion — to hold the boys in “involuntary servitude.” See *958United States v. Ancarola, 1 F. 676, 682-684 (CC SDNY 1880). Given the nature of the system the Padrone statute aimed to eliminate, the statute’s use of the words “involuntary servitude” demonstrates not that the statute was “aimed only at compulsion of service through physical or legal coercion,” ante, at 947, but that Congress understood “involuntary servitude” to cover servitude compelled through other means of coercion.5 Indeed, the official title of the Padrone statute was “An act to protect persons of foreign birth against forcible constraint or involuntary servitude,” Act of June 23, 1874, ch. 464,18 Stat. 251 (emphasis added); 2 Cong. Rec. 4443 (1874), and the legislative history describes the statute as broadly “intended to prevent the practice of enslaving, buying, selling, or using Italian children,” ibid. (Rep. Cessna) (emphasis added).6
*959In light of this legislative history, the Court of Appeals below concluded that § 1584 must at least be construed to criminalize nonphysical means of private coercion used to obtain the services of particularly vulnerable victims such as minors, immigrants, or the mentally disabled. 821 F. 2d 1186, 1190-1192 (CA6 1987). I agree with the Court, however, that this creation of specially protected classes of victims is both textually unsupported and inconsistent with Congress’ decision to eliminate such distinctions in enacting § 1584, ante, at 950, and thus turn to the task of defining what I regard as the proper construction of the statute.
II
Based on an analysis of the statutory language and legislative history similar to that set forth in Part I, the Government concludes that § 1584 criminalizes any conduct that intentionally coerces involuntary service. It is of course not easy to articulate when a person’s actions are “involuntary.” In some minimalist sense the laborer always has a choice no matter what the threat: the laborer can choose to work, or take a beating; work, or go to jail. We can all agree that these choices are so illegitimate that any decision to work is “involuntary.” But other coercive choices, even if physical or legal in nature, might present closer questions. Happily, our task is not to resolve the philosophical meaning of free will, but to determine what coercion Congress would have regarded as sufficient to deem any resulting labor “involuntary” within the meaning of § 1584.
The Government concludes that the statute encompasses any coercion that either leaves the victim with “no tolerable alternative” but to serve the defendant or deprives the victim of “the capacity for rational calculation.” Brief for United States 19, 33. As the Court notes, however, such a statutory construction potentially sweeps in a broad range of conduct that Congress could not have intended to criminalize. Ante, at 949. The Government attempts to avoid many of *960these problems by stressing that a victim does not lack “tolerable alternatives” when he simply has “no attractive or painless options”; the alternatives must be as bad for the victim as physical injury. Brief for United States 33. One can, however, imagine troublesome applications of that test, such as the employer who coerces an employee to remain at her job by threatening her with bad recommendations if she leaves, the religious leader who admonishes his adherents that unless they work for the church they will rot in hell, or the husband who relegates his wife to years of housework by threatening to seek custody of the children if she leaves. Surely being unable to work in one’s chosen field, suffering eternal damnation, or losing one’s children can be far worse than taking a beating, but are all these instances of involuntary servitude? The difficulty with the Government’s test is that although nonphysical forms of private coercion can indeed be as traumatic as physical force, their coercive impact is more highly individualized than that of physical and legal threats. I thus agree with the Court that criminal punishment cannot turn on a case-by-case assessment of whether the alternatives confronting an individual are sufficiently intolerable to render any continued service “involuntary.” Such an approach either renders the test hopelessly subjective (if it relies on the victim’s assessment of what is tolerable) or delegates open-ended authority to prosecutors and juries (if it relies on what a reasonable person would consider intolerable).7 Similarly, I agree with the Court that the difficulty of distinguishing the victim deprived of “the capacity for rational calculation” from the victim influenced by love, *961charisma, persuasive argument, or religious fervor is sufficiently great that the standard fails to define the criminal conduct with sufficient specificity.
The solution, however, lies not in ignoring those forms of coercion that are perhaps less universal in their effect than physical or legal coercion, but in focusing on the “slavelike” conditions of servitude Congress most clearly intended to eradicate. That the statute prohibits “involuntary servitude” rather than “involuntary service” provides no small insight into the central evil the statute unambiguously aimed to eliminate.8 For “servitude” generally denotes a relation of complete domination and lack of personal liberty resembling the conditions in which slaves were held prior to the Civil War. Thus, in 1910 and 1949, Webster’s defined “servitude” as the “[clondition of a slave; slavery; serfdom; bondage; state of compulsory subjection to a master. ... In French and English Colonies of the 17th and 18th centuries, the condition of transported or colonial laborers who, under contract or by custom, rendered service with temporary and limited loss of political and personal liberty.” Webster’s New International Dictionary of the English Language. And in 1913 and 1944 Funk and Wagnalls defined “servitude” as “[t]he condition of a slave; a state of subjection to a master or to arbitrary power of any kind” and cited the same colonial practice. Funk and Wagnalls, New Standard Dictionary of the *962English Language.9 Our cases have expressed the same understanding. “The word servitude is of larger meaning than slavery, as the latter is popularly understood in this country, and the obvious purpose was to forbid all shades and conditions of African slavery.” Slaughter-House Cases, 16 Wall. 36, 69 (1873). “[T]he term involuntary servitude was intended to cover those forms of compulsory labor akin to African slavery which in practical operation would tend to produce like undesirable results.” Butler v. Perry, 240 U. S. 328, 332 (1916). See also Bailey v. Alabama, 219 U. S. 219, 241 (1911); Hodges v. United States, 203 U. S. 1, 17 (1906).
I thus conclude that whatever irresolvable ambiguity there may be in determining (for forms of coercion less universal than physical or legal coercion) the degree of coercion Congress would have regarded as sufficient to render any resulting labor “involuntary” within the meaning of § 1584, Congress clearly intended to encompass coercion,of any form that actually succeeds in reducing the victim to a condition of servitude resembling that in which slaves were held before the Civil War.10 While no one factor is dispositive, complete *963domination over all aspects of the victim s hie, oppressive working and living conditions, and lack of pay or personal freedom are the hallmarks of that slavelike condition of servitude. Focusing on such a slavelike condition not only accords with the type of servitude Congress unambiguously intended to eliminate but also comports well with the policies behind the statute, for the concern that coerced laborers will be unable to relieve themselves from harsh work conditions by changing employers is less likely to be implicated if that laborer has a normal job with time off, personal freedom, and some money, and has contact with other people.11
This focus on the actual conditions of servitude also provides an objective benchmark by which to judge either the “intolerability” of alternatives or the victim’s capacity for “rational” thought: the alternatives can justifiably be deemed intolerable, or the victim can justifiably be deemed incapable of thinking rationally, if the victim actually felt compelled to live in a slavelike condition of servitude. True, in marginal cases it may well be difficult to determine whether a slave-like condition of servitude existed, but the ambiguity will be a matter of degree on a factual spectrum,12 not, as in the “no *964tolerable alternative” or “improper or wrongful conduct” tests, a matter of value on which one would expect wide variation among different prosecutors or jurors. The risk of selective or arbitrary enforcement is thus minimized, and the defendant who, as a result of intentional coercion, employs persons in conditions resembling slavery has fair notice regarding the applicability of the criminal laws. And many of the more troublesome applications of the Government’s open-ended test would be avoided. For example, § 1584 would not encompass a claim that a regime of religious indoctrination psychologically coerced adherents to work for the church unless it could also be shown that the adherents worked in a slavelike condition of servitude and (given the intent requirement) that the religious indoctrination was not motivated by a desire to spread sincerely held religious beliefs but rather by the intent to coerce adherents to labor in a slavelike condition of servitude.
This restrictive construction of limiting the statute to slavelike conditions, although necessary to comply with the rule of lenity given the inherent ambiguity of the statute *965where the coercion is neither physical nor legal, is not, however, necessary where the defendant compels involuntary service by the use or threat of legal or physical means. Because the coercive impact of legal or physical coercion is less individualized than other forms of coercion, we need be less concerned about selective or arbitrary enforcement; and the defendant who intentionally employs physical or legal means to coerce labor has fair notice his acts may be criminal. The ambiguity justifying a restrictive reading is, moreover, not present when the means of coercion are those at the heart of the institution of slavery, and it seems clear that Congress would have regarded a victim working under a legal or physical threat as serving in a condition of servitude, however limited in time or scope.13
III
In sum, I conclude that § 1584 reaches cases where the defendant intentionally coerced the victim’s labor by the use or threat of legal or physical means or the defendant intentionally coerced the victim into a slavelike condition of servitude by other forms of coercion or by rendering the defendant incapable of rational choice. I therefore concur in the judgment.
The District Court instructed the jury to incorporate the definition of “involuntary servitude” from § 1584 into 18 U. S. C. § 241. The parties did not challenge this incorporation either below or in this Court, but rather argued only that the § 1584 definition the District Court incorporated was incorrect. 821 F. 2d 1186, 1188, n. 3 (CA6 1987). I therefore believe it appropriate to address only the proper construction of § 1584. I note also that the § 241 count of the indictment charged a conspiracy to interfere with the “free exercise and enjoyment of the right and privilege secured to [the victims] by the Constitution and laws of the United States to be free from involuntary servitude as provided by the Thirteenth Amendment of the United States Constitution.” App. 177 (emphasis added). Thus, the parties may have assumed that § 1584 is a “la[w] of the United States” specifying the content of the constitutional right to be free from involuntary servitude, cf. ante, at 941, and that accordingly if respondents’ actions violated § 1584, the conspiracy to engage in those actions would necessarily constitute a violation of § 241. Such an assumption does not strike me as at all unreasonable. At any rate, for whatever reason the parties never raised the argument that the definition of “involuntary servitude” under § 241 should differ from that under § 1584, and I think it imprudent to decide that issue in the first instance in this Court and without briefing.
In other contexts, we have recognized that nonphysical coercion can induce involuntary action. For example, we have interpreted the federal crime of kidnaping to include the imposition of “an unlawful physical or mental restraint” to confine the victim against his will. Chatwin v. United States, 326 U. S. 455, 460 (1946) (emphasis added). Similarly, in determining when confessions are involuntary, we have noted “coercion can be mental as well physical.... [T]he efficiency of the rack and the thumbscrew can be matched, given the proper subject, by more sophisticated modes of‘persuasion.’” Blackburn v. Alabama, 361 U. S. 199, 206 (1960). “When a suspect speaks because he is overborne, it is immaterial whether he has been subjected to a physical or a mental ordeal.” Watts v. Indiana, 338 U. S. 49, 53 (1949) (plurality opinion of Frankfurter, J.).
Although not detailed by the Court, the Government introduced evidence that the Kozminskis (1) ripped a phone off the wall in the barn when one of the victims was caught using it, and did not simply “discourage” contact with relatives but falsely told relatives who asked to speak to the victims that the victims did not want to see them and falsely told the victims that their relatives were not interested in them; (2) falsely told neighbors that the victims were in their custody as wards of the State; and (3) refused to allow the victims to seek medical care, even when one was gored by a bull and the tip of the other’s thumb was cut off (both victims eventually became very ill while serving the Kozminskis). The.Court also neglects to mention that the Government has conceded that the victims were not forcibly held to work on the farm. 821 F. 2d, at 1188.
Because the Court today adopts an expansive but rather obscure understanding of what “physical” coercion encompasses, see nn. 5, 12, infra, it is difficult to tell which, if any, of the means of coercion described in the last two paragraphs the Court would deem “physical.”
The Court attempts to evade the inconsistency between its interpretation of § 1584 and the coercion covered by the Padrone statute by asserting that the child victims of the padrone system were in a “situatio[n] involving physical. . . coercion.” Ante, at 947. Yet the coercion involved, even as the Court describes it, was obviously psychological, social, and economic in nature: “These young children were literally stranded in large, hostile cities in a foreign country. They were given no education or other assistance toward self-sufficiency.” Ibid. Although it is heartening that the Court recognizes that strange environs and the lack of money, maturity, education, or family support can establish the coercion necessary for involuntary servitude, labeling such coercion “physical” is at best strained and (other than making the legislative history fit the Court’s statutory interpretation) accomplishes little but the elimination of whatever certainty the “physical or legal coercion” test would otherwise provide. See n. 12, infra.
The legislative history of the Slave Trade statute is less conclusive, but in explaining the necessity of reenacting this ban on importing slaves despite the abolition of slavery and without the statute’s original limitation to blacks, Senator Heyburn did make clear that the new statute was intended to protect those who come here “without being a party to the disposition of their services or the control of their rights, whether they be children of irresponsible years and conditions or whether they be people who, because of their environment or the condition of their lives, cannot protect themselves.” 42 Cong. Rec. 1115 (1908).
These problems are not solved by limiting the Government’s test to “improper or wrongful conduct that is intended to cause, and does cause, the other person to believe that he or she has no alternative but to perform the labor,” United States v. Mussry, 726 F. 2d 1448, 1453 (CA9 1984) (emphasis added), for the criminal has no way of knowing what conduct the prosecutor or jury will deem sufficiently improper or wrongful to criminalize.
Because, as a criminal statute, § 1584 must be interpreted to conform with special doctrines concerning notice, vagueness, and the rule of lenity, the issue here focuses on what central evil the words “involuntary servitude” unambiguously encompass in a way that can be defined with specificity. The interpretation of “involuntary servitude” here is thus necessarily narrower than it would be if the issue were what enforceable civil rights the Thirteenth Amendment provides of its own force or if the issue here concerned the scope of Congress’ Thirteenth Amendment authority to pass laws for abolishing all badges or incidents of slavery or servitude. See Jones v. Alfred H. Mayer Co., 392 U. S. 409, 437-444 (1968).
See also 9 Oxford English Dictionary 522 (1933) (“The condition of being a slave or serf, or of being the property of another person; absence of personal freedom. Often, and now usually, with the added notion of subjection to the necessity of excessive labor”); Webster’s American Dictionary of the English Language 1207 (1869) (“the state of voluntary or involuntary subjection to a master; service; the condition of a slave; slavery; bondage; hence, a state of slavish dependence”).
The case involving the crime of holding to slavery that is most contemporaneous with the 1948 passage of § 1584 defined a slave mainly in terms of total domination of person and services and lack of freedom. United States v. Ingalls, 73 F. Supp. 76, 78-79 (SD Cal. 1947).
Significantly, the Padrone statute, which encompassed coercion through other than physical or legal means, see supra, at 957-958, was designed to prevent boys from being “held in a condition of practical slavery,” 42 Cong. Rec. 1122 (1908) (Sen. Lodge), or “in something kindred to slavery,” 2 Cong. Rec. 2 (1873) (Sen. Sumner). See also United States v. Ancarola, 1 F. 676, 682-683 (CC SDNY 1880) (determining whether such boys were held to involuntary servitude by relying on the defendant’s control over the *963boys and his use of them for his profit and to the injury of their morals). These slavelike conditions can presumably be contrasted with the conditions normally implicated by “ ‘the right of parents and guardians to. the custody of their minor children or wards.’” Ante, at 944, quoting Robertson v. Baldwin, 165 U. S. 275, 282 (1897).
“The undoubted aim of the Thirteenth Amendment . . . was not merely to end slavery but to maintain a system of completely free and voluntary labor throughout the United States. , . . [I]n general the defense against oppressive hours, pay, working conditions, or treatment is the right to change employers. When the master can compel and the laborer cannot escape the obligation to go on, there is no power below to redress and no incentive above to relieve a harsh overlordship or unwholesome conditions of labor.” Pollock v. Williams, 322 U. S. 4, 17-18 (1944).
“That there may be marginal cases in which it is difficult to determine the side of the line on which a particular fact situation fall's is no sufficient reason to hold the language too ambiguous to define a criminal offense.” United States v. Petrilto, 332 U. S. 1, 7 (1947) (rejecting vagueness challenge to statute making it a crime to-coerce the employment of “persons in *964excess of the number of employees needed”). Ambiguity over such matters of degree is not obviated by the Court’s test, since it requires a determination of whether the degree of physical or legal coercion used was sufficient to compel “involuntary” service. Cf. Steward Machine Co. v. Davis, 301 U. S. 548, 590 (1937). Indeed, the Court introduces á far more profound uncertainty by adopting an understanding of “physical” coercion that encompasses a broad array of what might commonly be understood to be nonphysical forms of coercion. See n. 5, supra. Although these forms of coercion certainly deserve to be encompassed within § 1584, it is at best obscure under the Court’s test what line divides the forms of coercion that are covered by § 1584 from those that are not because the Court never defines its rather unique understanding of “physical” coercion. Instead, the Court seems to use “physical” as no more than a formal label it applies to those forms of coercion it deems sufficiently egregious to criminalize. Such a mode of analysis is, of course, conclusory. Worse, it merely reintroduces all the difficulties of the Government’s test in a more obscure and exacerbated form.
Like the Court, I put aside the exceptional cases it discusses ante, at 944.