Temporal expressions in historical texts

Temporal expressions in historical texts describe different temporal phenomena, such as a point in time or a time interval. A date can be described explicitly, e.g. 25 December 2015, or implicitly, e.g. Christmas 2015. In addition, there are relative expressions, such as ‛next year’, ‛today’, or even free-text phrases with temporal meaning, e.g. ‛after the engagement of Prince Harry’.

TimeML is a TEI-compatible specification language for annotation of temporal information in natural language. In 2009 it was approved as an international standard for time and event markup called ISO-TimeML. In TimeML each temporal expression can belong to one of the four types:

  • DATE for expressions describing a calendar date, e.g. 7 July 2015
  • TIME for expressions referring to a time of the day (e.g. midnight)
  • DURATION for expressions describing a duration (e.g. one week)
  • SET for expressions describing a set of times (e.g. each year)

Computational linguist Natalia Korchagina gathered for her dissertation on the subject ‛Temporal entity extraction from historical texts’, under the supervision of Prof. Dr. Martin Volk, temporal expressions from various sources. First, a collection of SSRQ articles from 1450 to 1550 were tagged by a rule-based temporal tagging system called HeidelTime and the automatically produced results manually verified by two human annotators afterwards. Additionally, temporal expressions were extracted from the SSRQ database of historical terms (keywords and normalised lemmas including spelling variants), Nachbarschaftsfällen from the Zurich Council protocols and legal proceedings from the 15th century (StAZH B VI 197–B VI 243), a list of saint feast days from Hermann Grotefends ‛Zeitrechnung des Deutschen Mittelalters und der Neuzeit’, as well as six digital editions of SSRQ including SG and ZH.

These lists containing ca. 3000 entries, labelled with one of the four types described above, will be used by histHub for building a thesaurus of temporal expressions as well as for the automatic tagging of source texts, and will be available to all interested parties.

Based on this corpus of temporal expressions it would be possible to explore, for instance, if the changing perception of time is reflected in language, or how the appearance of the mechanical clock and the Gregorian calendar in the Late Middle Ages affected people’s sense of time.

Illustration: the Ketzerturm with its clock face from the old plan of the city of Zurich, by Jos Murer, 1576.