Derrida suggests in his late essay Poetics and Politics of Witnessing that bearing witness must not essentially consist in proving, in confirming knowledge, in ensuring a theoretical certitude, a determinant judgement. It can only appeal to an act of faith (Derrida, 2005:191). “Bearing witness, if there is such a thing” (191), he argues, is event-like in nature, at root a pledge or a promise that something happened to me and before me. The for me of witnessing is thus not privative in kind but precisely the secret source of a provocation and of an opening toward others- “secret” here signifying the inaccessible singularity (but not indivisibility) of the witness, and of the moment of witnessing. “For me at least” marks the irreducible difference between the witness and those who are called upon to witness witnessing. To hear his or her testimony, which is to say those who are summoned to witness something to which, precisely, they cannot testify, namely that which appeared to the other (witness) as other. Witnesses to witnessing cannot be found, not as such, but his interdiction in no way prohibits witnesses from witnessing: as Derrida says, the “prohibition imposed on bearing witness” occurs “in the very place where one has to go on appealing to it”. (201).
All witnessing assumes as part of its structure that the addressee of the witness will never see what was once present to the witness. The addressee, the third, has no immediate access to what was witnessed. Of course, the witness is not present to what he or she recalls either; all he or she can do is attest now to his or her having-been-present. So the voice asks you to believe-“Believe me because I tell you to, because I ask it of you”-but to believe is never to subscribe to the conclusion of a syllogism. A testimony is a pledge of sworn faith, yet “You have to believe me” cannot be taken as a convincing demonstration. Derrida argues that witnessing appeals to the act of faith, which is always open to betrayal, infidelity and perjury. If perjury or false oaths were not at the heart of all witnessing-and there must always be the chance of perjury-there would be no testimony.
Derrida argues that “all responsible witnessing engages a poetic experience of language” (181).What is this poetic experience of language? For Derrida, it is the way that a testimony both reveals and conceals itself during the act of testimony. In a poem, specifically the work of Celan, it is the “possibility of a secret” which helps create the poetic experience. This possibility of the secret, for Derrida, is the possibility of the lie or perjury. However, in this essay, the possibility of the lie is part of the poetic experience of language. All statements are potentially testimony, in the sense that they are uttered with the intent to be believed, with the implicit air of truth. The possibility of the lie, which can be known only by the witness and not by the listener, is what marks the speech as testimony. The listener is always in the position of having to believe (or not) the testimony; it cannot be taken as “proof” or evidence. Derrida Protests throughout his essay that testimony is not proof. A testimony that is certifiable proof is no longer testimony. In addition to the possibility of the lie, there is the possibility that the witness is offering an untrue testimony in good faith. In his discussion of Plato and his tests in Dissemination, Derrida confirms that morality, truth, memory and dialects are the central domains of writing.  The witness could have a faulty memory or perception of the event. This does not make the testimony false, but it increases the poetic experience or “possibility of secrets” of the testimony.
Another part of this poetic experience is untranslatability. Poems are notoriously difficult to translate, due to the multiple meanings and the “possibilit[ies] of secrets” which they contain. As Derrida shows, testimony itself-both the word “testimony” and the act of witnessing” is untranslatable between languages. It also becomes, in a sense, an untranslatable experience. The witness relates a certain bodily, sensory experience, which has to be (imperfectly) translated through language. The listener cannot directly experience the event to which the witness testifies, so he or she must receive the translation of it in words. This difficulty of translation is part of the poetic experience of testimony. Visual or mathematical proof needs no translation-it is the non-proof of testimony that must be translated. The pitfalls of translation contain or obscure the multiple meanings, the possibilities of secrets that Derrida links to the poetic experience.
In discussion of aesthetics and emblematic of literary criticism associated with the Holocaust is a statement made by German social-cultural critic Theodor W. Adorno, “to write poetry after Auschwitz is barbaric” (Wu,1981, 34).  This famous yet controversial statement is often taken out of the original context. On the surface, the remark seems to suggest that poetry-writing should be suspended or stopped altogether after the Holocaust. Adorno’s assertion is invoked both to support and to refute arguments about the appropriateness of literary representations of the Holocaust and scholarly analysis about them. The larger context of his writing reveals this statement less as a categorical dictum against Holocaust literature than as an indication of a far-ranging crisis of representation, culture, ethics, philosophical systems, and knowledge after the holocaust. Adorno amplifies, “I do not want to soften my statement that it is barbaric to continue to write poetry after Auschwitz’s;…..literature must resist precisely this verdict, this is, be such that it does not surrender to cynicism merely by existing after Auschwitz” (1992:87-8). Like the central paradox that runs through much Holocaust literature- speaking the unspeakable, representing the unrepresentable- Adorno argues for a weighty literature that engages embattles cultural and ethical issues while acknowledging that the Holocaust may render such a literature impossible. 
Derrida’s “poetics and politics of witnessing” seems both most willing to exorcise and most prone to calling back into life as he engages Paul Celan’s poem and its undecidably singular and exemplary text of witness/ing. Celan’s poem, or at least those portions of it Derrida cites, is, as it emerges, a virtually ideal document from which Derrida can derive his poetics, primarily because of the tensions it sustains between the nominative particularity of its title (Aschenglorie”) and the departicularised abstractions of the lines that fall under that title and move it further from its apparent occasion of naming and address. A poem that seems simultaneously to demand and to refuse a discrete referent, “Aschenglorie”, contains within itself the twin and contradictory (and “impossible” and “dialectal”) imperatives of the in-itself and the for-itself, the particular and the conceptual, the singular and the exemplary. Derrida’s essay reproduces that undecidability-at least theoretically- as it adduces from Celan’s text a poetics of witnessing dually, impossibly, undecidably, and dialectally obligated both to the singularity of what Celan’s witness has witnessed and to the exemplarity of that act of bearing witness.
The last lines of Celan’s poem are “No one / bears witness for the / witness.”(201) Paul Celan was a Holocaust survivor and both his parents died at the hands of the Germans. Throughout his life Celan felt compelled to act as a poetic witness to the Holocaust and Derrida rightly points out this task is in itself impossible. He substantiates this with- “»¿That comes down to saying – always the same paradox, the same paradoxopoetic matrix – that as soon as it is guaranteed, certain as a theoretical proof , a testimony can no longer be guaranteed as testimony.” 
Derrida repeatedly cited Celan’s striking phrase, “Nobody bears witness for the witness”. One sees the force of what this means. The witness I bear, the testimony I give, can be given only by me alone. I alone can bear witness for what I witnessed. Witnessing is absolutely individual, sui generis, unique, private, singular. Derrida draws an extreme conclusion from this, namely that that no act of testimony can be verified. It is Derrida alone who knows that demand that arouses his respect and that he cannot and will not compromise. Nevertheless, as Derrida also reminds us, the word “testimony” comes from testis, from terstis, meaning the third. The witness testifies as third to some transaction between at least two others. However he also reminds us that “one cannot and should not (claim to) replace the witness of his own death, for example someone who perished in the hell of Auschwitz.” Although Derrida agrees “there is no witness for the witness” , he protests further on that “testimonies, will never, by definition, be proofs.