POST 1. THE SEMANTIC FIELD OF THE POLYP: DIGITAL EXPERIMENTS

Nadège Lechevrel, Bénédicte PercheronGisèle Seginger

The posts grouped under the heading ‘The polyp. Forms and knowledge’ aim to initiate a study of the epistemological context in which work and thought on the polyp have developed, in the  of science, literature, philosophy, historiography and politics. In the nineteenth century – the period on which the ANR/DFG BIOLOGRAPHES programme was focused – terms and representations flowed freely between disciplines. This research also required consideration at the outset of literary and scientific intertextuality, and we had to look back particularly to the eighteenth century, when the first scientific works on the animal nature of the polyp were published. New ideas were spread widely thanks to Diderot and d’Alembert’s Encyclopédie (which includes a ‘Polyp’ entry) and also to works of literature.

Because of the quantity and diversity of texts to be dealt with, as well as the terminological complexity involved – there are multiple terms (polyp, polypers, coral, zoophyte, hydra), and the word “polyp” is polysemic – we have used textometric tools (Google’s n-grams, TXM software) and visualisation tools, such as TreeCloud software, which makes it possible to create word clouds. Examining the corpus with these tools revealed key elements of our topic. The results thus obtained contributed to the formulation of useful hypotheses both for the periodisation and for an intertextual study of literary texts belonging to a time when knowledge was flowing across disciplines. These posts thus provide an example of how linguistic engineering (especially textometry), the history of science and literature can work together. They give an overview of the results of our analysis, which will also feed into the writing of a collective book and the papers at the final conference in the BIOLOGRAPHES project.

The experiment was carried out on digitised texts (in text format), accessible via online databases. It should be noted that automatic extraction and analysis required sorting, in a second stage, in order to exclude two other meanings of the word ‘polyp’, which also refers, on the one hand (in medicine), to an abnormal growth of mucous tissues and, on the other hand (in zoology and literature), to the octopus, in a usage that has existed from Pliny (Natural History, Book IX, XLVI) right down to the eighteenth century (Diderot and d’Alembert, L’Encyclopédie, 1765). While the medical meaning would never disappear, the use of the word ‘polyp’ to describe the octopus became rarer in the nineteenth century, although examples can still be found in the literature of Chateaubriand (Itinéraire de Paris à Jérusalem, 1811: 100) or Victor Hugo for whom the animal with its gigantic tentacles was a figure of the monstrous and the unspeakable.[1] But these cases are rare enough not to invalidate the automated analyses.

Before beginning our exploration of digital libraries and following the evolution of terms, as well as large-scale searches using n-grams (Note 4), we first defined the complex lexical field of the polyp using a more limited corpus. We thus began by selecting a set of representative texts. This corpus includes important and widely-circulated scientific texts, well-known texts of popularisation of the period – such as those of Louis Figuier, one of the great popularisers of the nineteenth century – as well as literary texts (Diderot, Chateaubriand, Flaubert, Balzac, etc.), which show that the knowledge about the nature of the polyp and the broader questions it raised as early as the eighteenth century had spread well beyond scientific circles. Indeed, this being with its hybrid nature displayed a strange vitality and a plastic power that fascinated not only scientists but also writers.

Based on the criteria set out above, we identified 54 scientific or literary works, which were entered into a spreadsheet. This spreadsheet provides access to the metadata concerning these works as well as the authors’ biographical information (obtained by sending SPARQL requests to the Bibliothèque Nationale de France: http://data.bnf.fr/sparql) or information about the text files (title, source, number of words, etc.). This spreadsheet will then be used to summarise the automated analyses.

Despite the efforts to restrict this initial corpus, this analysis encountered an immediate difficulty: the imperfections in the OCR scanning (optical character recognition) of certain open-access digital library texts, especially in the case of pre-19th century editions, which were almost impossible to digitise perfectly. The need to analyse texts that were as well scanned as possible determined which editions and sources were used but did not achieve flawless scanning.

The graphs below provide an overview of the current corpus, the modes of access, the distribution by type of material and the distribution according to chronology.

Fig. 1. Access modes and scanning quality

 

Fig. 1 – Access modes and scanning quality

Fig. 2 – Categories: scientific works, literary works, works of popularisation, correspondence

Fig. 3 – Histogram of the most represented decades in the corpus studied

Figure 1 shows the proportion of texts that we were able to use (63% of the referenced works) to conduct our textometric analyses and produce visualisations of the lexicon. The texts were pre-processed via the software, TXM  which labelled and lemmatised them[2].

This corpus then allowed us to study the co-occurrences of the ‘polyp’ form and to distinguish – thanks to two visualisations (fig. 4 and 5) in the form of word trees – co-occurring common nouns (N) from the co-occurring adjectives (ADJ).

Fig. 4 – Word cloud of co-occurring NOUN lemmas for the ‘polyp’ form

 

Fig. 5 – Word cloud of co-occurring ADJ lemmas for the ‘polyp’ form

These visualisations were then used either to form hypotheses about the representations of the polyp in texts of various kinds, or as a starting point for the development of lexical resources (lists or dictionaries of forms) relating to biological knowledge. Such resources, developed through repeated returns to the text, will in turn allow us to continue to study the flow of biological knowledge in nineteenth-century literature and to refine our interpretations.

 

Bibliography for this post

The editions listed are available online.

BALZAC, Honoré de, Physiologie du mariage [1829], Œuvres complètes de M. de Balzac, Paris, Furne, vol. 16, 1842-1848.

CHATEAUBRIAND, François-René de, Itinéraire de Paris à Jérusalem et de Jérusalem à Paris, Paris, Le Normant, 3 vols.

DIDEROT, Denis et D’ALEMBERT, Jean Le Rond, Encyclopédie ou Dictionnaire raisonné des sciences, des arts et des métiers, 1751- 1772, [ARTFL], 17 vols.

FIGUIER, Louis, La Vie et les œuvres des animaux : zoophytes et mollusques, Paris, L. Hachette, 1866.

FLAUBERT, Gustave, Charpentier, 1874.

HUGO, Victor, Les Misérables [1862], Paris, J. Lemonnoyer, G. Richard, 1890, 3 vols.

PLINE L’ANCIEN, Histoire naturelle, Dubochet, Le Chevalier et Cie, 1848-1850, 2 vols.

You can also read the following posts

Post 2: Polyp, zoophyte, hydra: a history of words

Post 3: The polyp in Diderot and Alembert’s Encyclopédie

Post 4: N-Grams and the interpretation of terminological change

Post 5: The literary polyp

Post 6: The polyp: a political image

Post 7: General Bibliography

[1] ‘These four men were not four men; they were a sort of mysterious robber with four heads, operating on a grand scale on Paris; they were that monstrous polyp of evil, which inhabits the crypt of society.’ (‘Ces quatre hommes n’étaient point quatre hommes c’était une sorte de mystérieux voleur à quatre têtes travaillant en grand sur Paris c’était le polype monstrueux du mal habitant la crypte de la société.’) (Victor Hugo, Les Misérables, 1890, Paris, J. Lemonnoyer, G. Richard, III, p. 280).

[2] TXM uses TreeTagger for lemmatisation. See Helmut Schmid, ‘Probabilistic Part-of-Speech Tagging Using Decision Trees’, Proceedings of International Conference on New Methods in Language Processing, 1994, pp. 44-49.


Laisser un commentaire

Votre adresse de messagerie ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.