Reader Comments

Post a new comment on this article

VU Amsterdam Metaphor Corpus

Posted by Gerard_Steen on 25 Mar 2014 at 08:47 GMT

The VU Amsterdam Metaphor Corpus offers a 186,000 word excerpt from the British National Corpus that has been manually annotated for all forms of expression of metaphor by means of a highly reliable method for metaphor identification that has been based on roughly the same theoretical background assumptions called MIPVU.

For further information, see
Steen, G.J. et al. (2010), A Method for Linguistic Metaphor identification (Amsterdam: John Benjamins),
http://www.metaphorlab.vu...
http://www2.let.vu.nl/oz/....

How does the output produced by the present algorithms compare with our findings?

No competing interests declared.

RE: VU Amsterdam Metaphor Corpus

yneuman replied to Gerard_Steen on 25 Mar 2014 at 11:44 GMT

Hi Gerard, I'm familiar with your work and even cited it in mine. The algorithms that we have designed aim to classify phrases into "literal" and "non-literal" categories. I must better understand what do you mean by "compare". Our algorithms aim to automatically determine whether a phrase is metaphorical and therefore can be compared to other algorithms addressing the same challenge. Please feel to correspond through my academic email. Best wishes Yair

No competing interests declared.