Meaning of chemical changes?

What are some linguistics projects that have used topic modeling to study historical changes in word meaning?

• http://en.wikipedia.org/wiki/Topic_modeling is a technique in for studying the statistics of word clusters in a corpus of texts. I'm not sure whether there are adequate historical corpuses to study word meaning change over time, but it's an interesting prospect. It may also be that the technique has not had a lot of uptake yet into academic linguistics (it is popular in technology fields).

• Answer:

  Historical linguistics is generally devoted to the study of languages which have very sparse written records for languages over very long periods of time. The focus of study is usually on word changes, given that it's based on known patterns of sound change. Statistical inference of semantics from texts as in topic modelling requires a great deal of text, e.g. large corpora, which as you suggest is somewhat lacking for distant past. However tracking meaning change is possible ever since mankind became extremely noisy in the written record. Hall, Jurafsky and Manning looked at data (see citation)  between 1978 and 2006. They actually got a bit 'meta' with this and looked at computational linguistics conferences to show their diversity of topics. Which isn't really the same as tracking changing in the meaning of a word. I would assume that to do this you would need to establish a semantic wordnet and try to show how semantic relations changed over time. I'd be extremely interested to know if anyone has done this (or if I'm of the mark on how this might be achieved). See: Hall, D., Jurafsky, D., & Manning, C. D. (2008, October). Studying the history of ideas using topic models. In Proceedings of the conference on empirical methods in natural language processing (pp. 363-371). Association for Computational Linguistics. Accessible here: http://web.stanford.edu/~jurafsky/hallemnlp08.pdf p.s. you might want to have a play with Google Scholar. You might uncover a treasure trove b using different keywords than I did.