Invited Talk: NLP and the Humanities: The Revival of an Old Liaison

Proceedings of the 12th Conference of the European Chapter of the ACL, pages 10–15, Athens, Greece, 30 March – 3 April 2009. c⃝2009 Association for Computational Linguistics NLP and the humanities: the revival of an old liaison Franciska de Jong University of Twente Enschede, The Netherlands fdejong@ewi.utwente.nl Abstract This paper present an overview of some emerging trends in the application of NLP in the domain of the so-called Digital Hu- manities and discusses the role and nature of metadata, the annotation layer that is so characteristic of documents that play a role in the scholarly practises of the humani- ties. It is explained how metadata are the key to the added value of techniques such as text and link mining, and an outline is given of what measures could be taken to increase the chances for a bright future for the old ties between NLP and the humani- ties. There is no data like metadata! 1 Introduction The humanities and the field of natural language processing (NLP) have always had common play- grounds. The liaison was never constrained to lin- guistics; also philosophical, philological and lit- erary studies have had their impact on NLP , and there have always been dedicated conferences and journals for the humanities and the NLP com- munity of which the journal Computers and the Humanities (1966-2004) is probably known best. Among the early ideas on how to use machines to do things with text that had been done manually for ages is the plan to build a concordance for an- cient literature, such as the works of St Thomas Aquinas (Schreibman et al., 2004). which was ex- pressed already in the late 1940s. Later on hu- manities researchers started thinking about novel tasks for machines, things that were not feasible without the power of computers, such as author- ship discovery. For NLP the units of process- ing gradually became more complex and shifted from the character level to units for which string processing is an insufficient basis. At some stage syntactic parsers and generators were seen as a method to prove the correctness of linguistic the- ories. Nowadays semantic layers can be analysed at much more complex levels of granularity. Not just phrases and sentences are processed, but also entire documents or even document collections in- cluding those involving multimodal features. And in addition to NLP for information carriers, also language-based interaction has grown into a ma- tured field, and applications in other domains than the humanities now seem more dominant. The impact of the wide range of functionalities that involve NLP in all kinds of information process- ing tasks is beyond what could be imagined 60 years ago and has given rise to the outreach of NLP in many domains, but during a long period the humanities were one of the few valuable play- grounds. Even though the humanities have been able to conduct NLP-empowered research that would have been impossible without the the early tools and resources already for many decades, the more …

Content truncated to 500 words