Stylometric analysis of Lafond's letter that never Wrote Simón Bolívar to San Martín

Recently, several documents have appeared that have been collected at different times, and many of them have been auctioned or donated to various institutions. For example, the letter signed by Simón Bolívar for the victory at the Battle of Cúcuta in 1813 was auctioned by Doyle in 2016 for $23,750 ¹. Another example is the 1521 letter by Hernán Cortés that was sold for $32,500 in 2017, and two years later, it was discovered that the letter was stolen from the Historical Archives of the Nation of Mexico ².

Other documents are also emerging that have subsequently been proven to be false, for economic reasons or even for public recognition. A famous example is Galileo's 1610 manuscript that was exhibited at the University of Michigan Library and later proved to be a forgery of Tobia Nicotra ³.

Therefore, it is necessary to develop computational methodologies that enable us to verify the authenticity of documents and to be able to rule out any counterfeiting with the help of information technologies. Thanks to this, it is possible to carry out linguistic studies using the tools generated by a discipline called stylometry, primarily dedicated to recognizing patterns in the written language ^{4, 5, 6}.

Stylometry is a discipline that began to be established by the Polish writer Wincenty Lutoslawski (1863–1954) to determine the chronology of the Plato Dialogues ⁷, as well as the author's discovery of an unknown comedy that was in the National Library of Spain and was recently attributed to Lope de Vega instead of Miguel Bermúdez, as indicated in the work file ⁸. Another example is the work of who have not been included as authors of the work ⁹.

Another example was the analysis of the Book of Mormon (wrote by Joseph Smith and Sidney Rigdon), where it is indicated that the author was not Joseph Smith ¹⁰, but later work suggested that it was none of them ¹¹. In light of the foregoing, the present paper examines the letter Lafond allegedly written by Libertador Simón Bolívar to General José de San Martín, in which the fate of the Republic of Ecuador was being conferred ¹².

In 1940, a letter from Simón Bolívar to General José de San Martín that was included in the book "San Martín y Bolívar en la entrevista de Guayaquil a la luz de nuevosdocumentosdefinitivos" by Eduardo Colombres Mármol suggested that the recently formed Republic of Ecuador be incorporated into either Colombia or Peru ¹². This letter is known as Lafond's letter. We will take into consideration the letter written by Simón Bolívar to General San Martin on 1822, which is one of the letters that were included in that book. It is interesting to note that there is not the original letter, only a transcription of it, and there is neither evidence of it nor any reference to it in other records of Bolívar.

Furthermore, various works based on Eduardo Colombres' book have established the veracity of the letters between the two Hispano-American liberators ^{13, 14}. Gabriel Pedro Mara Lafond de Lurcy (1801–1866), a Frenchman who served in the Peruvian navy, claimed to have met General San Martín on September 5, 1839, when he showed him his work relating to the memoirs of his travels to Paris, and then gave San Martín the letter to validate it, according to Lafond de Lurcy. According to José Pacífico Otero and Enrique de Gandía ¹⁴, other historians dispute this claim and contend that one of Bolivar's assistants actually delivered the letter.

Again, the validity of the Lafond letter is defended by the Peruvian historian Germán Leguía suggesting that Bolívar himself was lost to him, as did the Argentine historian Beatriz Bragoni who indicated its validity and even justified the exchange of letters between Lafond and General San Martín ¹⁴.

The Spanish version of the letter Lafond is attributed to Juan Bautista Alberdí (1810-1884) in his biography of General San Martín ¹⁵. In fact, on May 28, 1844, and subsequently on June 5, the letter was referred to in El Peruano after the work entitled "Bolívar y San Martín," where the letter would have been validated by General San Martín himself. However, the Bolivarian historian Vicente Lecuna as well as Cristóbal Mendoza asserted, without any doubt, that this letter is false ¹⁴, as were others that were in a bad book, with the sole purpose of distorting the image of Simón Bolívar by cataloging him as ambitious. To argue that, they indicated that the letter speaks, for example, of nineteen thousand Spanish veterans in Peru and it is false because that number was reached in 1824. Likewise, this letter is false because it refers to a request for assistance from San Martín to Bolívar, who, knowing the Libertador's trajectory, would have responded to the same.

Due to this, statistical methods based on stylometric studies will be used to confirm the veracity of this letter in the wake of the controversy caused by Lafond's letter. 

The letters of Simón Bolívar were obtained from the Center for Digital Humanities Virtual Library Miguel de Cervantes, which was created in 2000 to be a reference center for the relevant works of the Spanish language, available freely and free of charge from the University of Alicante (details at www.cervantesvirtual.com). From it, we obtained 16 letters from Simón Bolívar as detailed in Table 1

This table indicates the date and place from which it was written and to whom it is addressed. The selection criteria for the letters were random so as not to mislead the information. However, the transcription of the above-mentioned letters with the originals was not verified, suggesting that they are genuine.

Furthermore, eleven letters attributed to General San Martín (see details in Table 1) were included to verify that the methodology developed in the paper was able to discern between the two authors. The Lafond letter was obtained from documents that the Congress of the Republic of Peru brought to the public through its Office of Citizen Participation, available at (www.congreso.gob.pe/participacion/museo/congreso/mensajes/carta_jose_san_martin_a_simon_bolivar_reclamando_anexion_guayaquil_colombia_3_marzo_1822).

The next step was to create the linguistic corpus using the words from the various letters mentioned above. This allowed us to determine the frequency of the words used by each author because each author's usage of adjectives, substantives, and verbs is consistent across all of their writings, which is how we know who wrote a given document. Python programming was used to conduct these studies.

Tokenizing is the stylometric term for the division of texts into groups of words. It was simple to determine the frequency of appearance of words in texts once the words had been separated and their lengths were known ¹⁶. The current analysis is based on function words, or the distribution ratio of prepositions—a factor that is typically ignored when a document is forged. Recall that some of the propositions include, for instance, the words a, against, etc. Later, the delta function was established in accordance with Burrows' definition from 2002 ¹⁷, which is widely used as a linguistic measure that separates authorship from texts. Although it was previously thought to be just a quick way to establish who wrote a text, Burrows later confirmed that it is an excellent method for determining the authenticity of works ¹⁸.

The Delta function simply determines the frequency variation of the most common words in a text using a z-score ¹⁹ function. It should be borne in mind that z-scores are the difference between the relative frequency minus the average of the words in the corpus, divided by the standard deviation of such data ^{details in 19}.

From these results, a matrix is constructed where the different frequencies obtained on each letter are compared, calculating the distance to Manhattan ²⁰. Such a matrix is usually visualized with the help of a dendrogram, that is, a hierarchical representation of the tree type, where the results are grouped into clusters, where texts that are related to each other, i.e., of the same author, appear in the same nodes of the branches.

Figure 1 displays the results of the frequency of the words derived from Simón Bolívar’s letters (shown as columns in this figure), while the results derived from Lafond's letter are shown as a black line. This graph demonstrates that there is no proportion of functional words, among which are highlighted "la," "a," "no," "los," "me," "lo," "este," "le," and "esta," indicating that the Lafond letter does not reproduce the style of the Liberator's Bolívar.

Figure 1.It shows the distances determined by the word frequencies of Simón Bolívar’s letters (all of which begin with the letter B for easy identification), as well as the distance determined by the “evil letter” of Lafond (the result is shown as a black line).

Figure 2 shows the result represented as a dendrogram (Figure 2). This figure shows that the letters are forming clusters grouped by the same author; that is, it is presented how only the letters of Simón Bolívar or of General San Martín are grouped among themselves, noting that both authors have their own style when writing.

Moreover, the Lafond letter is not included in any cluster formed by other letters from any of the liberators of Hispano-American; that is, the letter does not reproduce the linguistic footprint of the Libertador Bolívar, much less the style of San Martín. Thirdly, the style of writing of San Martín has little variation, unlike the richness of style present in the letters of Bolívar, because they are grouped into different branches.

Figure 2.It shows the rectangular dendrogram that was produced by stylometric analysis of the letters written by Simón Bolívar, San Martin, and Lafond. San Martín's letter begin with SM, while Bolívar's letter begin with the B.

The goal of this paper is to find out the validity of Lafond's letter through stylometric analysis and to determine whether Simón Bolívar was really disposing of the future of the Republic of Ecuador with General San Martín. After analyzing the Delta function, it was found that Simón Bolívar never wrote the aforementioned letter to General San Martín. Therefore, it is confirmed that Lafond's letter is false and only sought to damage the reputation of Libertador Bolívar.

1. 1.RPP Redacción. (2016) 22) "Carta que firmó Simón Bolívar en las guerras de Independencia fue subastada". Available at https://rpp.pe/mundo/actualidad/carta-que-firmo-simon-bolivar-en-las-guerras-de-independencia-fue-subastada-noticia-1011778 (Retrieved.
   Search at Google Scholar

1. 2.Villagran L. (2020) 2) "Allegedly stolen from Mexico, letters from Spanish conquest go for thousands at auction" Available at https://www.elpasotimes.com/story/news/2020/10/02/hernan-cortes-letters-stolen-mexico-sell-new-york-inah-search/3520423001/ (Retrieved.
   Search at Google Scholar

1. 3.Feldman E. (2022) 22) "Historian Discovers a Prized Galileo Manuscript Was Forged" Available at https://www.smithsonianmag.com/smart-news/historian-discovers-a-prized-galileo-manuscript-was-forged-180980615/ (Retrieved.
   Search at Google Scholar

1. 4.Bensalem I, Paolo Rosso, Chikhi S. (2019) On the use of character n-grams as the only intrinsic evidence of plagiarism". Language Resources and Evaluation. 53(3), 363-396.
   Search at Google Scholar

1. 5.Stamatatos E. (2009) A survey of modern authorship attribution methods. , JASIST 60(3), 538-556.
   Search at Google Scholar

1. 6.Fuller S, O'Sullivan J. (2017) Structure over Style: Collaborative Authorship and the Revival of Literary Capitalism. , Digital Humanities Quarterly 11(1).
   Search at Google Scholar

1. 7.Lutoslawski W. (1898) Principes de stylométrie appliqués à la chronologie des œuvres de Platon". Revue des Études Grecques. 11(41), 61-81.
   View article·Search at Google Scholar

1. 8.Bvmc. (2014) 23) "Descubierta una comedia inédita de Lope de Vega" Available at https://blog.cervantesvirtual.com/descubierta-una-comedia-inedita-de-lope-de-vega/ (Retrieved.
   Search at Google Scholar

1. 9.Schoenbaum S.Internal evidence and Elizabethan dramatic authorship; an essay in literary history and method, Editorial.
   Search at Google Scholar

1. 10.Jockers M L, Witten D M, Criddle C S. (2008) Reassessing authorship of the Book of Mormon using delta and nearest shrunken centroid classification”. , Literaty and Lingustic Computing 23(4), 465-491.
   Search at Google Scholar

1. 11.Schaalje G B, Fields P J, Roper M, Gregory L. (2011) Extended nearest shrunken centroid classification: a new method for open-set authorship attribution of texts of various sizes”. Digital Scholarship in the Humanuties 26(1), 71-88.
   Search at Google Scholar

1. 12.Mármol Colombres, Eduardo. (1940) San Martín y Bolívar en la Entrevista de Guayaquil a la luz de nuevos documentos definitivos, prólogo de Rómulo Carbia, Buenos Aires, Coni. , n° 10.
   Search at Google Scholar

1. 13.Bragoni Beatriz. (2016) El intercambio epistolary entre San Martín y Lafond. , Prismas 20(1), 47-60.
   Search at Google Scholar

1. 14.Perez Amuchástegui AJ. (1962) La carta de Lafond. Cordoba : Universidad Nacional de Cordoba, Instituto de Estudios Americanistas. Edicions Siglo XX, Buenos Aires.
   Search at Google Scholar

1. 15.Alberdí J M.(1844) "Biografia del Jeneral San Martín" Editrial Imprenta de Ducessois , Quai Des Augustins.
   Search at Google Scholar

1. 16.Škorić M, Stanković R, Ikonić Nešić M, Byszuk J, Eder M. (2022) Parallel Stylometric Document Embeddings with Deep Learning Based Language Models in Literary Authorship Attribution. Mathematics. 10-5.
   Search at Google Scholar

1. 17.Burrows J F. (2002) Delta: A Measure of Stylistic Difference and a Guide to Likely Authorship. , Literary and Linguistic Computing 17(3), 267-287.
   Search at Google Scholar

1. 18.Stefan E. (2017) others, Understanding and explaining Delta measures for authorship attribution. Digital Scholarship in the Humanities, Volume 32, Issue suppl_2 4-16.
   Search at Google Scholar

1. 19.Patergianakis A, Limniotis K. (2022) Privacy Issues in Stylometric Methods. Cryptography. 6-2.
   Search at Google Scholar

1. 20.Stanikunas D, Mandravickaite J, Krilavicius. (2017) Comparison of distance and similarity measures for stylometric analysis of Lithuanian texts. CEUR Workshop proceedings [electronic resource]: ICYRIME , Kaunas, Lithuania Vol, 1852.
   Search at Google Scholar