Eld P Csirmaz

The Purloined Reader

On a Possible Model of Embedded Narrations


An apparent difference can be perceived between the analyses of Poe’s “The Purloined Letter” by Lacan and by Derrida, namely, the treatment of the topmost (outermost) frames of narration. Derrida even accuses Lacan of equaling the narrating Narrator with the narrated Narrator and/or the Inscriber:


But what the Seminar treats is only the content of this story, what is justifiably called its history, what is recounted in the account, the internal and narrated face of the narration. Not the narration itself. […] One might be led to believe, at a given moment, that Lacan is preparing to take into account the (narrating) narration [and] the very curious place of the narrator. But once it is glimpsed, the analytic deciphering […] neutralizes it [which] transforms the entire Seminar into an analysis fascinated by a content. (179)


This essay aims at providing an alternative to both the views of Derrida and the analysis of Lacan by pushing Lacan’s description of the structure of the first scene of Poe’s story to the extreme, suggesting that the (number of) frames in which a given piece of information appears do not alter the reader’s interpretation of it; and by attempting to give a (theoretical) definition of the various beings or entities that participate in the creation of narrations mentioned by Derrida in the light of the model of communication described in Jakobson’s essay “Linguistics and poetics”—which consideration, in turn, leads to similar consequences than those of the first argument.

That Lacan, at many points, appears to hold the opinion that information retransmitted by a further entity, a further narrator is usually not distorted fundamentally—which idea is equal to the view that all narrators are equal to the Author—is suggested by many remarks throughout his essay. “The narration, in fact, doubles the drama with a commentary” (29) he suggests at the beginning, and when he describes Dupin’s quotations, witty remarks and anecdotes, he writes “No doubt Poe is having a good time” (37). That is, it is Poe who is having a good time, not Dupin, which means that according to Lacan, they are either equivalent, or at least they have the same opinion.

Lacan also describes the phenomenon of retransmission on a theoretical level. He appears to argue that the fact that the first scene (when the letter is purloined by the Minister and lost by the Queen) is narrated by a series of narrators (first, by the Queen, then, by the Prefect, then, by the (narrating) Narrator and finally, by the Inscriber and/or the Author) without fundamental alterations proves that the scene “belongs to the dimension of language” (35). He also adds: “Thus the indirect telling sifts out the linguistic dimension [from the whole scene], and the general narrator, by duplicating it, ‘hypothetically’ adds nothing to it” (35). In other words, the narrations of the first scene work according to the suggestions of this paper, that is, they do not alter the original information by relating it again.

This is the suggestion that I attempt to generalize to all narrations of Poe’s story. This view is in contrast with Lacan’s analysis, who contrasts the two purloining scenes by suggesting that the second one (narrated by Dupin) persists in the “symbolic order” (34), and that in it, the role of the general narrator is entirely different (35). Still, my suggestion is not entirely inconceivable taking into account that it is in the analysis of the second part of the story that Lacan equals Dupin with the Author by the above-quoted remark.

My suggestion is also in an apparent contrast with Derrida’s analysis, as if the information is not altered by retransmission, then the retransmitting agents, the various entities described by Derrida (the narrating and narrated Narrators, the Inscriber and the Author) are unimportant, transparent, and thus, nonexistent for the Reader of the story. Derrida, holding the opposite view, even introduces these entities into the ‘triads’ determined by Lacan, suggesting that “there is always the supplement of a square whose opening complicates the calculations” (179), which immediately entails that following his analysis, the retransmitting agents cannot be regarded as nonexistent.

Let me add at this point, that the two suggestions are reconcilable, as even if it is supposed that the entities do not alter the messages they retransmit, there might be messages originating in them, which fact renders them existing—and able to be introduced into the network of figures of the story.


In order to support the alternative suggestion that messages are not altered fundamentally by retransmission, let us turn to Poe’s text and investigate whether parts related to the Reader via different series of narrators show considerably different characteristics. I chose to investigate the lexical level, as it is on the level of style that character-based differences appear to be the most easy to introduce. However, it was found that on this level no differences whatsoever could be found.

After preparing a list of all words in the story, those words were selected for further investigation which


(1) are concrete enough to carry some stylistic effect; function words were excluded;

(2) appear in the text more times;

(3) are not related solely to one segment of the text (for example, to the description of the map-puzzle or the child prodigy) as then, their use is restricted to the speaker who relates the given segment;

(4) preferably have a synonym with approximately the same distribution so that character-based differences could manifest in the preference for one alternative over the other;

(5) and if (4) does not hold, they are not key words of the story itself.


It was based on this (rather subjectively determined) set of words that I found that Poe employs little, if any, stylistic difference to distinguish between the speakers.

Although the absence of lexical differences is impossible to prove based on examples, let me recount the distribution of some words to illustrate their universal usage in Poe’s story regardless of the speaker who utters them. The indication of the series of narrators via which the word appears is based on the following abbreviations: A=Author; N=narrating Narrator or Narrator-present, which here, is regarded as equal to the Inscriber; N0=narrated Narrator or Narrator-past, the figure in the story; D=Dupin; D0=Dupin-past; P=Prefect; P0=Prefect-past; Q=Queen; Q0=Queen-past. The speaker of the word (who is quoted directly) is typeset in boldface. After the speaker, a further chain of narrators might be noted in parentheses which indicate the source of the information. For example, the story of the first purloining is related via A_N_P(Q_Q0), the narration of the second purloining can be represented as A_N_D(D0).






“we carefully probed”

A_N_P(P0) [about the search]

“examined it carefully”

A_N [on D taking the check]

“which I had carefully prepared”

A_N_D(D0) [on the fake letter]


(a possible synonym, hide, is not even used in the text)

“her wish to conceal it”

A_N_P(Q_Q0) [first scene]

“paper can be concealed”

A_N_P(P0) [on search]

“he may have concealed it”

A_N_N0 [on Minister]

“the person wishing to conceal an article”

A_N_P [on possible hiding places]

“all men proceed to conceal a letter”

A_N_D [on Prefect’s thinking]

“a disposal of the article concealed”


“in searches for articles concealed”

A_N_D [on Minister]

“to conceal this letter”

A_N_D [on Minister]

“not attempting to conceal it at all”

A_N_D [on Minister]


(plan is not used in the text; the distribution of idea is similarly balanced)

“he must design in the end”

A_N_P [on Minister]

“giving him reason to suspect our design”

A_N_P [on Minister]

“a certain rich miser conceived the design”

A_N_D() [the Abernethy-story]

“to which he forcibly adapts his designs”

A_N_D [on Prefect]

“as if a design, in the first instance, to tear it”

A_N_D(D0) [on the letter found]

“so suggestive of a design to delude”



The list of words with balanced usage could be continued with certain (used 1 times by N, 5 times by P, 6 times by D), document (as a synonym for letter and paper, 5P, 4D, 1N), exactly (1P, 2D), fool(s) (2P, 4D), individual (3P, 2D), ingenious (1P, 2D), letter(s) (1 in the title, 8P, 8N0, 1N, 17D), matter (1N, 2N0, 2P, 8D), paper(s) (3P, 3N0, 4D), possession (2P, 3N0, 2D), premises (2P, 3N0, 4D), etc. As can be seen from the balanced usage of word groups like letter–paper–document, none of the three main speakers (and not even the narrating Narrator) has a distinct style, and as the usage of sophisticated words like premises, possession, ingenious, individual shows, this, common style is close to that of highly educated people, or of those who want to be seen as such. It is also interesting that the words ha, ho and oh, which are characteristic of the Prefect only, and thus may establish a distinct style for him (although their use is too marginal to achieve that effect) is carefully balanced by the puffs of Dupin.

The absence of stylistic differences may show that the real origin of all messages is the Author or the Inscriber (here, the Narrator-present) and the real addressee of all passages of the text (an unsurprising suggestion) is the Reader, while other, formal narrators permit the message to flow through them without altering it. They may, however, attach a logical tag on the message, as apparently, the Reader is aware that a given sentence expresses the Prefect’s or the Narrator’s opinion, but this tag, in my view, does not alter the message itself. It might negate it, if the Reader is led to believe that the message comes from an unreliable source (the Reader, most probably, does not believe that the Prefect’s team really has looked everywhere—which suggestion shows why there is no real contradiction arising from the fact that Dupin finds the letter on the premises, which has been thoroughly searched, as suggested by Lacan (38)), but cannot change its internal structure.

This suggestion appears to hold more for the outermost layers of the narration (the distinction between Author, Inscriber, narrating and narrated Narrator) than for the figures themselves, as if there really were no distinction between messages coming from different figures, dramatic texts, for example, would be uninterpretable. Therefore, let me restrict the suggestion that narrations, that is, retransmissions do not alter messages fundamentally for embedded or subordinated, not coordinated sources. This way, our suggestion is that it is hard, if not impossible, to distinguish between fiction, narration and narrated narration; between the Queen’s story and the Prefect’s rendition of it; between Dupin’s opinion and those of Poe himself; but it is easy to separate the arguments of the Prefect and of Dupin. The fact that distinctions between entities of narration, even between Author and Narrator were introduced, it appears, rather late in the history of criticism also supports the idea that the usual method of the Reader is to neutralize embedded sources.


Having suggested the hardly separable nature of the narration-creating entities described by Derrida, let me venture into an even more theoretical consideration of a possible definition for these entities based on the model of communication described by Jakobson.

According to Jakobson, any verbal communication has six main factors: addresser, message, addressee, context, contact and code (35). Having determined that in the cases of the entities of the Author, the Inscriber and the Narrators, we are dealing with embedded narrations, such model of these narrations shall be proposed (restricting it for two levels at the present moment) which consist of a main communication, the message of which contains an embedded addresser, message, addressee, context, contact and code. From now on, the factors of the main communication shall be written uppercase.



Let us try to give, based on this model, an account for the relationship between the Author and the narrating Narrator. I have excluded, from this consideration, the Inscriber, who may be the source of the motto and the title of Poe’s story, provided it is to be upheld that it is still not the Author who put those segments around the narration. The Inscriber, however, can be easily reintroduced into the model to be proposed by inserting a further layer between the Author and the Narrator.

In our model, the most general addresser, the ADDRESSER, is equaled to the Author. The ADDRESSEE, who is ‘outside’ the text, just like the Author–ADDRESSER, is the Reader. CONTEXT may refer to the historico-cultural environment in which the work appeared, while, unsurprisingly, the MESSAGE is the work or text itself. CODE, as always, refers to a common language. CONTACT might be hard to determine, as most artistic verbal communications happen in written, that is, in a deferred form. It might also be argued that such texts rarely contain phatic (or, rather, PHATIC) messages, although the summary chapters in Henry Fielding’s Tom Jones might be considered as such passages. Also, it can be supposed that CODE=code in all embedded communications, as otherwise, the message would not be decipherable for the ADDRESSEE, who ‘overhears’ the message formally intended for addressee.

The addresser can be regarded to be the Author-in-text, generated solely by the text (MESSAGE) itself, thus it is the (narrating) Narrator, the lyrical I, etc. The message is the narration, as opposed to the work (fiction), which is the MESSAGE. According to the model, context is the environment according to which message is interpreted. It might be more than the CONTEXT if it is presupposed that the title, the motto and other related texts are not parts of the narration, the message, but parts of the MESSAGE—which forces them to enrich the context.[1] We have arrived at the Reader-in-text, the addressee, whose identification proves to be problematic. Whenever the narrator (addresser) addresses the Reader, it might be supposed that the (formal) addressee of the message is addressee. However, if no such direct (and conative) message can be found in a text, the placement of its addressee is still problematic. Based on the observation that the ADDRESSER (Author) is usually identified with the addresser (Narrator) in a text, it might be attempted to determine what the Reader identifies with while receiving the MESSAGE. In most of the cases, this is the Narrator, the addresser, hence we get the addresser=addressee formula as the general property of the top two embedded layers of the narration of artistic texts. (The addresser is what the ADDRESSER is identified with, hence it is the Narrator; the addressee is what the ADDRESSEE is identified with, hence it is again the Narrator.)

Moreover, it can also be argued that the real addressee of all messages (the MESSAGE, the message and the further embedded ones) is the Reader, hence addressee=ADDRESSEE. Which one is the more probable suggestion is hard to determine: the placement of the addressee appears to depend greatly on the structure of the work. It is based on the fact that while the notion of the addresser has been introduced many times, under the names of the narrator, the lyrical I, the agent, etc. but the identification of the addressee, whose existence is supposed based on Jakobson’s model, has not been addressed extensively, in fact, it appears to be missing from artistic texts that I presumed that in the creation of artistic texts, the Reader, or, rather, the Reader-in-text is ‘purloined.’

Please note that not all embedded communications are addressee-less. If a quotation is embedded into the narration, i.e. a further level can be found inside the message, then the embedded addressee, formally, is the figure who the embedded message is addressed to, and really—as it is overheard by both the addressee and the ADDRESSEE, even by the addresser and the ADDRESSER—is the Reader.

Following Roland Barthes’s suggestion, that the Author should not be taken into account when analyzing a literary text and in fact, it is by the disappearance of the Author that a text becomes artistic (170), it can be suggested that the MESSAGE should be stripped of its environment, of the ADDRESSER, the ADDRESSEE, the CONTACT, of course, and even, as some critical schools advocate, the CONTEXT. If so, it might not be surprising that the agents of the message, the addresser and the addressee, having no available antecedent, are turned on, and identified with each other. The self-reflexive nature of artistic messages, as suggested by Jakobson (37) also supports the reduction of their environments.


Finally, let me attempt to relate the embedded Jakobsonian model of artistic communication to the model of composition and reception established based on the views of va Babits. This model distinguishes between three layers of artistic works. On layer I resides the objective text; layer II hosts all structures built up by the text and on layer III the interpretation of the text by the Reader and the Author’s original experience can be found. It is also supposed that on layer II, in most cases, a representation of the Author can be found, which is generally termed the lyrical I. Based on Babits’s views, I have suggested disregarding layer III in literary analyses. This action is similar to the above suggested ‘stripping’ of the MESSAGE from its communicative environment, as the ADDRESSER and the ADDRESSEE, the interpretations of whom exist on layer III, can themselves be regarded to be on layer III. The narration, the message, may be layer I itself,[2] and its communicative environment, existing solely in the MESSAGE, may provide the salient elements of layer II (addresser=lyrical I, for example). The ‘base of associations’ in the model based on Babits’s views, which is shared by both the Author and the Reader and which ensures that the Reader will have similar associations than the Author, can be a complex of the CODE and the CONTEXT.

According to these parallels, on the figure representing two embedded narrations, it is outside the larger rectangle that layer III can be found; layer II resides between the two rectangles, and layer I is ‘in’ the innermost rectangle—it is the narration itself.


In conclusion, let me emphasize the parallel between the two main arguments presented in this essay. It has been argued that the suggestion is not unsupportable that embedded narrations have no direct effects on the structures of messages: they will surface and get to the Reader regardless of how many times they have been retransmitted. We have also found that the addressees of embedded communications, most importantly, the addressee, whose existence is predicated based on the Jakobsonian model of communication, are usually not defined in any way, and it was unprecedentedly hard to find some meaningful interpretation for them. A possible reason for this is that they do not exist, that the addressee of embedded messages is always the Reader, which entails that embedded communications are not communications on their own: they are reduced versions of full communicational schemes that try to behave as if they were embedded.

Suggesting that embedded communications merely appear to be embedded leads to the idea that messages are not retransmitted, in the Lacanian sense; they are simply pretended to be retransmitted, in which case it is not unexpected to find that they are not fundamentally transformed in this nonprocess.



Works Cited


Babits, va. General lectures on literary theory. Mihly Fazekas Secondary School, 1997–2001.

Barthes, Roland. “The Death of the Author.” Modern Criticism and Theory: A Reader. Ed. D. Lodge. London: Longman, 1988. 167–171.

Derrida, Jacques. “The Purveyor of Truth.” Trans. Alan Bass. The Purloined Poe: Lacan, Derrida & Psychoanalytic Reading. Eds. John P. Muller & William J. Richardson. Baltimore & London: Johns Hopkins UP, 1988. 173–212.

Jakobson, Roman. “Linguistics and Poetics.” Modern Criticism and Theory: A Reader. Ed. D. Lodge. London: Longman, 1988. 32–57.

Lacan, Jacques. “Seminar on ‘The Purloined Letter’.” Trans. Jeffrey Mehlman. The Purloined Poe: Lacan, Derrida & Psychoanalytic Reading. Eds. John P. Muller & William J. Richardson. Baltimore & London: Johns Hopkins UP, 1988. 28–54.

Poe, Edgar Allan. “The Purloined Letter.” In The Complete Illustrated Stories and Poems. London: Chancellor, 1994. 319–333.

[1] Provided there is no internal level for the Inscriber, whose message would contain these elements.

[2] This statement is complicated, however, by the distinction between elements which are present in the text, and thus, are on layer I, and elements which are merely associated with them, which still could be regarded parts of the message.