Multilingual FrameNet Project

The spread of English FrameNet and the growth of many projects creating FrameNets for other languages (see FrameNets in other languages) mean that we can now start to ask really significant, large-scale questions, such as:

  • To what extent are the semantic frames developed for English appropriate for other languages? Are some frames universal?
  • Are there certain semantic domains in which frames tend to vary more across languages?
  • Are there regular patterns of differences based on language families, regional groupings, etc.?

The FrameNet team has received a grant from the National Science Foundation (#1629989) to make such research possible, by building, in collaboration with the teams from other languages, a database of alignments of frames and frame elements across languages. The team at ICSI is now working on preparing the data from all the languages and defining the alignment algorithms.

In addition to providing their current data for alignment, teams from many languages have agreed to annotate a number of texts for which translations are available across languages; this will allow direct cross-linguistic comparisons of specific uses of words. The first such text will be the transcriptions of the very popular TED talk "Do schools kill creativity?".