Petition for review granted; vacated and remanded by published opinion. Judge GREGORY wrote the opinion, in which Judge FLOYD and Judge THACKER joined.
GREGORY, Circuit Judge:
Faustin Mukadi Ilunga, a citizen of the Democratic Republic of the Congo, appeals the denial of his application for asylum and protection under the Convention Against Torture (“CAT”). We hold that the rejection of Ilunga’s asylum application, largely on the basis of an adverse credibility finding, was not supported by substantial evidence. We thus remand for further proceedings.
I.
The following description of Ilunga’s travails in the Congo and his journey to the United States is based on his asylum application, testimony before the Immigration Judge (“U”), and corroborating documentation in the record. The IJ’s adverse credibility determination necessarily called into question the trustworthiness of many of the facts alleged.
A.
Before fleeing to the United States-, Ilunga lived in the Congo with'his wife and five children. In 2003, he joined the Movement for the Liberation of the Congo (“MLC”), a political party that actively opposed President Joseph Kabila in the country’s 2006 elections. Ilunga was a paid employee and- member of the party, participating in highly visible campaign activities and public appearances in the city of Lubumbashi.
After the MLC candidate lost the 2006 election to President Kabila, Ilunga’s political activism endangered him. Local police and others loyal to President Kabila threatened Ilunga’s life and vandalized his home. The police also killed two MLC supporters with whom Ilunga worked during the campaign. Increasingly fearful, Ilunga wrote a letter to his childhood friend living in neighboring Zambia, Bernard Kabeya, expressing his anxiety while accusing the president of assassinating his father.
The letter was intercepted by government agents working for the Congolese intelligence agency, the Agence Nationale de Renseignements (“ANR”). On December 23, 2006, an undercover ANR agent went to Ilunga’s home, blindfolded him, and drove him to prison where he was interrogated. Ilunga admitted that he authored the letter, and the ANR agent stated that Ilunga “would be killed” as a result. A.R. 61.
The government sent Ilunga to prison where he spent more than a month in a small cell shared with Jean Nkongolo Raíala. Ilunga suffered daily torture. Prison guards stabbed him and poured battery acid in the wounds. They shocked him with an electrical club, routinely whipped him, and raped him.
On February 2, 2007, Ilunga and Raíala escaped from prison with the help of a guard whom they paid off. The pair fled to Zambia in the bed of a truck hauling copper. While Ilunga remained in Zambia, the government tortured his family, raped his wife, and burned his home.
On June 22, 2008, Ilunga and Raíala boarded a plane for the United States. Ilunga’s wife and children fled to a Zambian refugee camp.
B.
Ilunga arrived at Washington Dulles International Airport without a visa. He told an immigration officer that he “left [his] country for political reasons” and was “looking for asylum.” A.R.2006. He further specified that he was “afraid to go back home” and had “no doubt” that he would be harmed again if he returned to the Congo. A.R.2006.
At a credibility hearing three weeks later, Ilunga attested to his party membership, the threats against him as a result of his political activity, the circumstances surrounding his arrest, the torture he endured in prison, and his escape. The asylum officer determined that Ilunga established a credible fear of persecution.
C.
In May 2009, Ilunga filed his application for asylum, withholding of removal, and CAT protection. The application is consistent with the account he gave at the credibility hearing, and it provides greater detail about the arrest and torture in prison, including incidents when Ilunga was stabbed, shocked with an electrical device on the genitals, sexually assaulted, and beaten.
For supporting documentation, Ilunga provided:
• His affidavit, detailing his abuse and escape;
• A medical affidavit from Dr. Michael Viola, who examined Ilunga in the United States and found that: (1) Ilunga’s “reporting of his torture history and symptoms are notable because of his consistent and precise description of specific details and the correlation of his history to his present symptoms and physical findings”; (2) Ilunga’s psychological symptoms are consistent with moderate post-traumatic stress disorder (“PTSD”); (3) Ilunga’s physical injuries are consistent with cuts with a sharp object, and his chest wound is consistent with “delayed healing of a wound secondary to infection or the reported pouring of
acid into the cut”; and (4) Ilunga’s fear of return to the Congo is credible;
• An affidavit from Kalala, Ilunga’s cellmate, that is consistent with Ilunga’s statements but does not speak to the specific torture that Ilunga suffered in prison;
• An affidavit from Bernard Kabeya, Ilunga’s friend in Zambia, who confirmed Ilunga’s account of his time in that country;
• Extensive documentation of country conditions in the Congo, including descriptions of politically-motivated violence, state-sponsored executions, forced disappearances, torture in prison, and impunity for rape;
• Ilunga’s reissued MLC membership card and a letter from the MLC expressing concern “about his survival” and attesting that his “activism on behalf of democracy in our country has caused him a lot of trouble from the security officers who are in power (ANR)”;
• Photographs purporting to show scars on Ilunga’s body caused by his torture in prison;
• Letters from Ilunga’s wife and family warning about conditions in the Congo and detailing their flight to Zambia; and
• The refugee card, and registration attestation, issued to Ilunga’s wife by the United Nations High Commissioner for Refugees (“UNHCR”).
D.
Ilunga and Kalala testified at Ilunga’s removal hearing. Ilunga’s primary language is Tshiluba, but he claimed to speak French fluently, and a contract French interpreter translated the proceedings. After reviewing the two days of testimony and the record, the IJ found that Ilunga was not credible and denied his application for asylum, withholding of removal, and CAT relief.
Three pieces of testimony were central to the Id’s credibility determination. First, the IJ cited supposedly inconsistent statements about the location of Ilunga’s torture inside the prison, and whether Kalala witnessed it. Ilunga, the IJ observed, stated that he was beaten away from his cell and that only guards witnessed the beatings. In seeming incongruity, Kalala’s translated testimony provides that “these things took place in the same room where they spent their nights” and that Kalala “was there” when guards stabbed Ilunga. The IJ determined that such an inconsistency “cannot be explained by a translation error, particularly as one claimed to have witnessed the beatings in the same room in which they slept while the other testified to have been taken to a different part of the prison and to have been beaten with only guards present.”
Second, the IJ found inconsistencies in testimony regarding the prayer practices of Ilunga and Kalala. Ilunga, the IJ determined, was initially non-responsive when asked exactly how the cellmates prayed together, and he was vague when pressed about the timing and content of the prayers. Moreover, the IJ claimed Ilunga was “hesitant and vague” when answering questions about the frequency and timing of prayer.
Third, the IJ agreed with the government that there were material inconsistencies in the dates on the MLC documentation provided. The date on the membership card and the letter, the IJ determined, was the day before Ilunga’s detention, even though Ilunga claimed that his wife obtained the documents much later while he was living in the United States. Moreover, the IJ believed it was significant that the MLC letter did
not mention Ilunga’s arrest, “which would be expected if it had been written when claimed and if there was a typographical error in the date.”
Regarding the testimony as a whole, the IJ found that Ilunga’s demeanor also supported the adverse credibility finding, noting in a single sentence that Ilunga was non-responsive and appeared uncomfortable answering some questions. The IJ added in passing that “DHS has also raised some valid concerns regarding plausibility and vagueness.”
Finding Ilunga incredible, the IJ determined, “necessarily calls into question all aspects of [his] claim.” The IJ further concluded in two sentences that Ilunga was not entitled to asylum based on any independent evidence unrelated to those credibility findings.
As for Ilunga’s application for CAT relief, the IJ separately held that “there [was] not sufficient reliable, independent evidence in the record to demonstrate that it is more likely than not that the Respondent would be tortured if he had to go back to the Democratic Republic of the Congo.”
E.
On July 25, 2013, the Board of Immigration Appeals (“BIA”) upheld each of the IJ’s determinations, largely adopting the IJ’s factual findings and reasoning. Specifically, the BIA credited the IJ’s account of testimonial inconsistencies, deferred to her demeanor observations, and agreed with her assessment of the documentary evidence. The BIA additionally affirmed the denial of CAT relief.
II.
The Immigration and Nationality Act (“INA”) gives the Attorney General the discretionary power “to grant asylum to aliens who qualify as ‘refugees.’ ”
Dankam v. Gonzales,
495 F.3d 113, 115 (4th Cir.2007). To qualify, applicants must establish they are unable or unwilling to return to their country of nationality because of past persecution “or a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group or political opinion.” 8 U.S.C. § 1101(a)(42);
see also id.
§ 1231(b)(3). An applicant can establish eligibility simply by providing credible testimony about his or her experiences. 8 C.F.R. § 208.13(a).
When the BIA affirms and adopts an IJ’s decision, this Court reviews both decisions as the final agency action.
Marynenka v. Holder,
592 F.3d 594, 600 (4th Cir.2010). An asylum order must be upheld unless “manifestly contrary to the law and an abuse of discretion.” 8 U.S.C. § 1252(b)(4)(D). Our review of an adverse credibility determination is specifically limited to ensuring that substantial evidence supports it.
See Qing Hua Lin v. Holder,
736 F.3d 343, 351 (4th Cir.2013) (citing
Dankam,
495 F.3d at 119). Although broad deference extends to the agency’s determination, it “must provide specific, cogent reasons” supporting its decision.
Djadjou v. Holder,
662 F.3d 265, 273 (4th Cir.2011).
The REAL ID Act of 2005 requires that credibility determinations be based on the totality of the circumstances, including:
the demeanor, candor, or responsiveness of the applicant or witness, the inherent plausibility of the applicant’s or witness’s account, the consistency between the applicant’s or witness’s written and oral statements (whenever made and whether or not under oath, and considering the circumstances under which the
statements were made), the internal consistency of each such statement, the consistency of such statements with other evidence of record ... and any inaccuracies or falsehoods in such statements, without regard to whether an inconsistency, inaccuracy, or falsehood goes to the heart of the applicant’s claim.
8 U.S.C. § 1158(b)(l)(B)(iii). As this Court has held, “omissions, inconsistent statements, contradictory evidence, and inherently improbable testimony are appropriate bases for making an adverse credibility determination.”
Djadjou,
662 F.3d at 273.
The totality of the circumstances standard thus provides an IJ with ample discretion in assessing credibility. It does not, however, permit a judge to “cherry pick” facts or inconsistencies to support an adverse credibility finding that is unsupported by the record as a whole.
Ai Jun Zhi v. Holder,
751 F.3d 1088, 1091 (9th Cir.2014) (quoting
Shrestha v. Holder,
590 F.3d 1034, 1040 (9th Cir.2010));
see also Hanaj v. Gonzales,
446 F.3d 694, 700 (7th Cir.2006) (“The IJ cannot selectively examine evidence.... ”);
Shah v. Att’y Gen. of U.S.,
446 F.3d 429, 437 (3d Cir.2006) (same). Instead, an IJ’s determination must take into account all the evidence submitted, including independent documentary support.
See Shah,
446 F.3d at 437. Thus, at a minimum the IJ must consider the petitioner’s explanation for any inconsistency to verify that an inconsistency actually exists, and then evaluate whether the discrepancy renders the entire testimony incredible in light of the record as a whole.
See Shrestha,
590 F.3d at 1044.
As described above, the IJ based her credibility determination on four grounds: (1) testimony about the location of Ilunga’s torture in prison; (2) testimony about the prayer practices of Ilunga and Kalala inside their cell; (3) the dates on the MLC membership card and letter obtained by Ilunga; and (4) Ilunga’s demeanor during testimony. We consider each in turn.
1.
When asked where he was “whipped” in prison, Ilunga first testified that it was “inside the cell” and then stated he did not understand a follow-up question about whether he was “in the cell” with his cellmate at the time. Asked to elaborate, Ilunga’s translated reply was: “Being as they called, they take one out. They whipped you good and they take you back inside.” Minutes later, when asked why he had told a doctor that he was attacked outside his cell, Ilunga stated “[the abuse] was inside the prison but not in the cell where I was staying in.” He again struggled to understand follow-up questions about whether he was additionally tortured “inside” the “cell.”
Kalala, meanwhile, testified that “he was there” when Ilunga was tortured, and that the abuse “was in the same room where we spent our nights.” On re-direct, in front of a different translator, Kalala stated that the first translator had used the French word for “jailhouse” instead of “room”, leading to any discrepancy between his testimony and that of Ilunga.
A single testimonial discrepancy, particularly when supported by other facts in the record, may be sufficient to find an applicant incredible in some circumstances.
See Djadjou,
662 F.3d at 275 (affirming an adverse credibility finding when the applicant had testified she was in hiding, but an eviction notice indicated that she was in her store on the same day). But such an inconsistency provides inadequate justification when, as here, there is a strong indi
cation it results from translation errors or language-based misunderstanding, particularly when it is belied by an extensive record of otherwise consistent statements and corroborating evidence.
See Shrestha,
590 F.3d at 1040 (observing that an IJ may not “cherry pick solely facts favoring an adverse credibility determination while ignoring facts that undermine that result”).
As an initial matter, the trustworthiness of any translation, when at issue, must be a relevant factor in the agency’s analysis of the totality of the circumstances supporting a credibility finding. Simply put, there must be reasonable assurances that any inconsistencies in testimony are, in fact, real and not the product of interpretation errors, language-based confusion, or similar factors.
See Perez-Lastor v. INS,
208 F.3d 773, 778 (9th Cir.2000) (“[A]n incorrect or incomplete translation is the functional equivalent of no translation: the alien must be able to understand the questions posed to him and to communicate his answers to the IJ.”).
As is well established, a defective translation of an immigration proceeding can amount to a constitutional due process violation when: (1) the plaintiff can show he or she did not “receive a full and fair hearing on [his or her] claims,”
Rusu v. INS,
296 F.3d 316, 321-22 (4th Cir.2002), and (2) the violation caused prejudice such that the results of the proceedings were likely impacted,
id.
at 324 (citing
Farrokhi v. INS,
900 F.2d 697, 703 n. 7 (4th Cir.1990)). But even absent a constitutional violation, “faulty or unreliable translations can undermine the evidence on which an adverse credibility determination is based.”
He v. Ashcroft,
328 F.3d 593, 598 (9th Cir.2003) (citing
Balasubramanrim v. INS,
143 F.3d 157, 162-64 (3d Cir.1998)).
Determining whether a flawed interpretation fatally undermines a credibility determination requires an examination of direct and indirect evidence .of error. The Eighth and Ninth Circuits have sensibly identified three types of evidence that “tend to prove that a translation was incompetent.”
Perez-Lastor,
208 F.3d at 778;
Tun v. Gonzales,
485 F.3d 1014, 1029-30 (8th Cir.2007) (adopting the
Perez-Lastor
framework). As
Perez-Lastor
established:
First, direct evidence of incorrectly translated words is persuasive evidence of an incompetent translation. Second, unresponsive answers by the witness provide circumstantial evidence of translation problems. A third indicator of an incompetent translation is the witness’s expression of difficulty understanding what is said to him.
208 F.3d at 778 (citations omitted).
Ilunga’s hearing transcript bears all three hallmarks of unreliability. Before turning to the specific testimony regarding the location of torture in the prison, it is instructive to examine the hearing as a whole. The two days of testimony were translated by two different interpreters, with nearly all instances of potential confusion arising on the first day. The transcript of that day reveals:
• An instance when the interpreter failed to translate Ilunga’s statement that he was sexually assaulted in jail, an omission caught by Ilunga’s attorney, leading to the translator’s opaque admission after the fact that he thought Ilunga “was not speaking clearly and this interpreter, perhaps, what he could hear from him [sic]” (A.R.615);
• A repeated disconnect between questions and answers
(see
A.R. 598 (Ilunga’s attorney asked about threats made, and Ilunga responded with a statement about damage done in a store); A.R. 609 (the IJ asked about
how Ilunga was arrested, and Ilunga replied that he had been told that two of his colleagues had been killed, and that he was afraid); A.R. 612-13 (Ilunga’s counsel asked Ilunga to describe his prison, and Ilunga answered “I do not know” and provided other disconnected answers); A.R. 616 (Ilunga’s attorney asked where his largest scar was, and Ilunga pointed to his chest but answered it was “on the right arm and on the right knee”); A.R. 639 (the government asked about where Ilunga’s wife obtained an MLC letter, and Ilunga replied “[b]ecause our party quarters are located in Kinshasa”); A.R. 661 (the government asked where an event took place, and Kalala answered with when it took place));
• At least 16 times during the first day’s testimony when Ilunga or Kalala stated they did not understand a question
(see
A.R. 592 (twice), 595, 604 (three times), 608, 609, 631, 636, 638, 641, 644, 646, 652, 653);
• An instance when Ilunga’s attorney believed the interpreter translated Ilunga’s statement as saying he suffered at the hands of the political party he worked for, instead of the party he opposed (A.R.606);
• Confusion about whether Ilunga testified that he told his wife and children to flee their home while he was still in prison or after (A.R.634);
• At least 11 times when the interpreter needed a question repeated
(see
A.R. 585, 591, 593, 597, 602, 605-06, 609, 610, 621, 640, 643, 650); and
• Frequent grammatical errors and questionable word choices
(see, e.g.,
A.R. 585, 597-98).
Those red flags should inspire special caution before an IJ parses translated statements to assess inconsistencies.
Indeed, here it is impossible to say that Ilunga’s testimony was inconsistent with that of Kalala regarding the location of torture because we do not know precisely what the men testified to. Instead, the transcript reveals that the cellmates were consistently, and genuinely, confused about the questions regarding location. As one example, the IJ observed that Kalala stated he saw the torture inflicted on Ilunga, in apparent tension with Ilunga’s statement that he was beaten in a different part of the prison with only guards present. Kalala, however, was only translated as stating that he was “there” when the stabbing and beatings took place. Thus, even taking the translation on its face, it is unclear whether Kalala testified that he “saw” the abuse or not. Furthermore, according to Kalala’s statement during the second day of testimony, the first translator used a French word that connotes “jailhouse” instead of “room” when describing the cell in question, causing confusion in his testimony.
See
A.R. 717. Neither the IJ nor the BIA resolved which word was used, or whether Kalala’s claim was correct. Instead, the BIA observed that Kalala nonetheless testified that Ilunga “was beaten in the 10 square foot room where they both slept.” Kalala, however, is only translated as stating that he himself had been beaten in that “room,” not that Ilunga was ever beaten there.
Against that backdrop, the IJ’s reliance on the alleged testimonial inconsistency was unfounded.
2.
The IJ also based her credibility determination on asserted inconsistencies in testimony regarding the prayer practices of Ilunga and Kalala in prison. The IJ specifically determined that Ilunga was initially non-responsive and then vague when asked about his prayer practices. Moreover, the IJ found he was “hesitant and vague” when asked about the frequency and timing of his prayers. . .
Tellingly, the BIA did not treat the prayer testimony as inconsistent, but rather as part of the IJ’s demeanor assessment. For good reason. The transcript reveals that there was not a single substantive inconsistency between the testimony of Ilunga and Kalala. Both testified they prayed together. Both testified they knelt to pray. And both testified they prayed for their release from jail. Nothing more was asked of them.
Moreover, any hesitancy and vagueness cited by the IJ is consistent with the repeated disconnect between questions and answers throughout the proceeding— strong indirect evidence of interpretation problems. The relevant portion of the transcript is as follows:
Q: How often did you and he pray together?
A: I cannot say how many times.
Q: Okay. My question is how often? How many times a day did you pray with him?
A: Every time.
Q: And exactly how did you and he pray together?
A: First, after my arrival in the prison, I did not know him and I was afraid of him. I did not want to know who he was. And the day I was cut on the knee, that’s when I started praying in my native language, in Tshiluba. And he heard me praying. Then he said we are the same. We are coming from the same region. Then he explained to me where he came from. That’s when we decided to start praying together.
Q: And my question was, exactly how did you and he pray together?
A: I don’t know.
Q: You don’t know how you and he prayed together?
A: We kneel down and we pray in our native tongue.
Such a labored exchange, when considered in the context of the entire transcript, says more about communication failures than it does about Ilunga’s credibility in answering the questions.
See Tun,
485 F.3d at 1031 (observing that language-based difficulties in understanding can lend “an air of evasiveness and confusion to the proceedings”).
The agency’s credibility determination, however, again failed to consider the quality of the interpretation. And given the otherwise consistent nature of the substantive testimony regarding prayer practices, it was an abuse of discretion to use such testimony to find Ilunga incredible.
3.
The IJ further cited Ilunga’s MLC membership card and letter from the party as supporting an adverse credibility finding. Both are dated December 24, 2006, the day after Ilunga’s arrest.
Bunga, however, testified that he asked his wife to obtain the documents after he arrived in the United States in 2008. Such a discrepancy, the IJ concluded, constituted a “material inconsistency].”
The BIA, however, failed to meaningfully consider Bunga’s reasonable explanation for the apparent inconsistency. According to Bunga, MLC officials dated .the documents to show no lapse in membership after the ANR ripped up his original card when he was arrested. Bunga openly acknowledged in his affidavit and live testimony that the card in the record was a reissued replacement, and that his wife obtained it to help with his asylum claim. Moreover, his description of the destruction of his original card at the hands of security officials is consistent with the documented practice “for government officials to illegally arrest MLC party members and confiscate their MLC party cards.” Br. of Appellants 31 (citing Amnesty Int’l,
Democratic Republic of Congo: Torture and Killings by State Security Agents Still Endemic
328 (October 2007) (included in the record at A.R. 386)).
In rejecting Bunga’s account, the BIA merely observed that he failed ‘to make his explanation before the IJ, but the BIA cited no authority for why it could ‘not consider the explanation on appeal. We conclude that any ambiguity that may exist about the date on the card and letter is insufficient to sustain an adverse credibility determination given Bunga’s plausible explanation, the agency’s conclusory treatment of it, the absence of any contrary evidence, and the extensive record corroborating Bunga’s claim. The IJ’s presumption that MLC officials should have dated the documents when Bunga’s wife requested them amounts to speculation and conjecture.
See Ayi v. Gonzales,
460 F.3d 876, 883-84 (7th Cir.2006) (rejecting an IJ’s “speculative leap” and determination that a party membership card was forged based on a forensics analysis that showed a “paper disturbance” on the card);
Zuh v. Mukasey,
547 F.3d 504, 510 (4th Cir.2008) (questioning “the appropriateness of speculating about foreign documents” (citing
Ayi,
460 F.3d at 883));
Lin-Jian v. Gonzales,
489 F.3d 182, 189 (4th Cir.2007) (observing that “we will not defer to an adverse credibility finding that is based on speculation, conjecture, or an otherwise unsupported personal opinion” (quoting
Tewabe v. Gonzales,
446 F.3d 533, 538 (4th Cir.2006) (internal quotation marks omitted))).
The IJ, however, also reasoned that if the letter had been written when Bunga claimed, it should have mentioned Bunga’s arrest. As we have previously observed, letters 'written by political parties attesting to an individual’s political involvement need not mention such arrests to be credible.
Tassi v. Holder,
660 F.3d 710, 724 (4th Cir.2011) (finding unpersuasive the government’s argument that letters describing an individual’s political activities should have mentioned arrests). In this case, Bunga obtained the document to corroborate his political involvement with the party, not as evidence of his arrest. It is entirely speculative to suggest that the party official should have both known of Bunga’s arrest and included it in the letter. Furthermore, the letter does provide further independent confirmation that Bunga suffered persecution at the hands of the state on account of his political activi
ties. It specifically expresses concern “about [Ilunga’s] survival” and attests that his “activism on behalf of democracy in our country has caused him a lot of trouble from the security officers who are in power.... It is in this way that he has been threatened many times, searched and intimidated during the presidential elections of 2006.”
The agency’s reliance on the MLC documentation to support an adverse credibility determination was thus also unfounded given the record as a whole.
4.
Finally, Ilunga urges this Court to reject the IJ’s demeanor-based findings because the IJ failed to ground her conclusions in specific facts. The IJ offered two principal demeanor observations: (1) Ilunga and Kalala “appeared uncomfortable when asked detailed questions concerning their claimed time together;” and (2) Ilunga “appeared non-responsive at times and uncomfortable answering some questions.” Such statements echo those made regarding the prayer practice testimony. Ilunga maintains that such broad-brush statements are insufficient to provide a specific, cogent ground for an adverse credibility determination.
An inherent tension exists in evaluating an IJ’s demeanor-based conclusions in asylum proceedings. On the one hand, broad deference understandably extends to a judge who is in the best position to gauge the demeanor of a witness and the presentation of testimony.
See Rusu,
296 F.3d at 323. On the other hand, linguistic and cultural differences, combined with the effects of trauma, caution against normative determinations.
See Dia v. Ashcroft,
353 F.3d 228, 274, 277 n. 6 (3d Cir.2003) (en banc) (McKee, J., concurring in part and dissenting in part) (sounding caution about demeanor assessments based on cultural norms, particularly for those who have been traumatized). Furthermore, as previously observed, difficulties in understanding during an asylum hearing can also lend “an air of evasiveness and confusion to the proceedings.”
Tun,
485 F.3d at 1031.
In affirming the IJ, the BIA summarily disagreed with Ilunga’s argument that it was normal for a victim of torture to appear “uncomfortable” given his experiences. The BIA’s disagreement manifests a basic misunderstanding of the human condition. In this ease, the record suggests that Ilunga was subjected to a pattern of vicious abuse, leaving both body and mind scarred by the experience. As Dr. Viola diagnosed, Ilunga suffered from moderate PTSD as a result of his experiences. Forced to revisit that trauma at the immigration hearing, Ilunga specifically testified about being raped by prison guards and subjected to other forms of sexual abuse. The record also indicates that he cried while testifying about the torture he endured. For the BIA to dismiss the potential impact of such torture on Ilunga’s testimonial disposition is unsettling.
Indeed, the ability to testify in a cool and collected manner about an experi
ence of torture would arguably raise greater credibility concerns.
Finally, the IJ cited no specific behavior or mannerisms that gave her pause. Instead, she merely stated that Ilunga appeared “uncomfortable.” Such a conclusion fails to provide a “specific, cogent reason[ ]” supporting a credibility determination, particularly given both the aforementioned interpretation issues and the nature of the testimony at issue.
See Djadjou,
662 F.3d at 273.
Even if his testimony was incredible, we additionally find that the IJ failed to sufficiently consider whether Ilunga presented adequate independent documentary evidence to establish asylum eligibility. As we held in
Camara v. Ashcroft,
independent evidence may establish past persecution on a protected ground even if an IJ finds the victim’s testimony to be incredible. 378 F.3d 361, 370-71 (4th Cir.2004);
see also Djadjou,
662 F.3d at 275. When actual past persecution can be shown, “a presumption arises that [the applicant] has the requisite level of fear of persecution, and thus she need not prove the subjective component of ‘well-founded fear.’ ”
Camara,
378 F.3d at 369-70 (quoting 8 C.F.R. § 208.13(b)(1)).
The central question here is thus whether the non-testimonial evidence independently established that Ilunga suffered persecution as a result of his political activities. Such evidence need not include a “smoking gun” or direct proof of persecution on account of political opinion.
See INS v. Elias-Zacarias,
502 U.S. 478, 483, 112 S.Ct. 812, 117 L.Ed.2d 38 (1992) (observing that an asylum applicant must provide some “direct or circumstantial” evidence of a persecutor’s motives). Instead, an applicant may meet his or her burden by presenting a consistent body of circumstantial evidence.
See id.
Here, the strongest pieces of such independent evidence are: (1) the doctor’s report obtained in the United States that concludes Ilunga’s wounds are consistent with the torture he described; (2) the MLC membership card and letter that expresses concern “about his survival” and links Ilunga’s political activity with his suffering “a lot of trouble from the security officers who are in power”; (3) the UNHCR refugee card and registration attestation issued to Ilunga’s wife, corroborating Ilunga’s statements that she was forced to flee the Congo; (4) photographs, including those of Ilunga’s scarred body and burned house; (5) Ilunga’s passport showing he left the Congo for Zambia before entering the United States; and (6) extensive documentation of country conditions, describing pervasive violence against minority political parties and activists.
Confronted with that body of evidence, the IJ discounted the MLC documentation during her credibility analysis for the reasons described above. She also stated that the medical affidavit “does not prove what caused the medical issues noted” and that additional letters from friends and family “do not overcome the credibility concerns.”
We agree with the IJ that absent the MLC membership card and letter, thfere is insufficient independent evidence in the record to support Ilunga’s asylum claim. But in light of our determination that the IJ improperly discredited the MLC documentation, the agency should consider on remand whether the documents, when combined with the other circumstantial evidence in the record, establish that Ilunga was a member of the MLC, was active in the party, and was persecuted as a result of his political opinions.
See Li v. Gonzales,
405 F.3d 171, 177 (4th Cir.2005) (observing that “[persecution involves the infliction or threat of death, torture, or
injury to one’s person or freedom, on account of one of the enumerated grounds”).
III.
We thus grant Ilunga’s petition for review insofar as it challenges the denial of his application for asylum, and we vacate the BIA and IJ’s orders with regard thereto.
We remand the case to the BIA for further proceedings consistent with this opinion. If the BIA chooses to further remand the matter to an IJ, we recommend that it schedule the case before a different judge.
PETITION FOR REVIEW GRANTED; VACATED AND REMANDED.