AI identifies 16th century lost work by Spanish Golden Age author

Spain’s National Library has been using artificial intelligence to transcribe anonymous historic literary works. In one of its recent finds, a previously unknown play was found to be written by one of the nation’s greatest authors Felix Lope de Vega. Named “La francesa Laura” (The Frenchwoman Laura), the play was written a few years before his death in 1635. 

This discovery was made by Etso, a project started by Golden Age literature experts Germán Vega and Álvaro Cuéllar in 2017, as reported by The Guardian. This project uses AI to determine the authors of anonymous or misattributed Golden Age plays. The Golden Age of Spanish literature extended roughly between the early 16th century and the late 17th century. Nicknamed Monstruo de Naturaleza (Monster of Nature), a Golden Age dramatist and playwright, Lope has written 500 plays, three novels, and several thousand sonnets during his lifetime. Some of his popular works include "Fuente Ovejuna" and "The Knight from Olmedo".

As part of Etso, 1,300 plays were digitally transcribed by Transkribus, an AI-based text recognition and transcription tool for historical documents. Transkribus managed to identify 3 million words from the plays studied. Post transcription, the plays were compared for language and style with the digitised work by 350 authors in the Etso database. At this point, the play (La francesa Laura) matched with over 100 works by Lope. 

As per the leaders of the Etso program, this is a significant feat because the play had close to nil references in any of the bibliographies that they came across in the past. If not for AI, the play would have remained unknown, they added. Four years back, the Etso program was able to identify a 17th-century play written by Mexican dramatist Juan Ruiz de Alarcón. 

AI is being increasingly used to recognise historic texts and literature.  Enrique Jiménez, a professor of Ancient Near Eastern Literatures at Ludwig Maximilian University in Germany, along with his team has been processing Babylonian literature using AI since 2018. Till now, the team has managed to process 22,000 text fragments.

In 2022, researchers from AI research firm DeepMind, Oxford University, and others, developed Ithaca, a deep neural network for restoring and attributing ancient Greek inscriptions. The authors of the study said that the tool was able to achieve 62% accuracy in restoring damaged texts and 71% accuracy in attributing it to its original geographical location.