I want to automatically create a list of words for Charades (in German or English), where one person has to describe, draw or show by pantomime a certain word to the others. Humans can intuitively choose words that are neither too easy nor too difficult, that are possible to characterize with the chosen method (especially pantomime is really hard for some words) and that are not boring. However I need help to (1) to define those criteria formally so that I can implement it in a program and (2) to find lexical databases or word lists with all the necessary attributes. The following are my ideas so far.
- I assume: Some POS tags are better than others. Verbs are good for pantomime. Nouns are good in general. Concrete things are easier than abstract concepts. Words like "of", "and" and "from" are hard.
- assume difficulty is correlated to popularity
- approximate popularity as: occurrence rank in a corpus, number of links to Wikipedia article
- Wikipedia article/DBpedia resource name (does not provide any linguistic information but good category system and ontology)
- WordNet has POS tags but no rank
- http://corpora.uni-leipzig.de has corpora with has rank but no POS tags