Emprical Approach in Cognitive Linguistics

Computational Cognitive Linguistics

Cognitive linguistics now needs empirical evidence and more computational basis to prove own 'linguistics and semantics' claims by Lakoff and Langacker and move ahead to the next step.

Some of starting points here:

Josh Tenenbaum and his group's research
Holyoak's recent work (especially [1] or [2])
Shimon Edelman. Bridging language with the rest of cognition: computational, algorithmic and neurobiological issues and methods. in the Proceedings of the Ithaca EMCL workshop. (John Benjamins, 2007).
Jerome Feldman. Computational Cognitive Linguistics. COLING 2004: Proceedings of the 20th International Conference on Computational Linguistics. 2004.
Camara Pereira, Francisco. Creativity and Artificial Intelligence: a conceptual blending approach. MOUTON DE GRUYTER. December 2007.

I am wondering recently that a challenge in computational cognitive linguistics might emerge from connections to machine learning and corpus linguistics which could give statistical and theoretical validity to cognitive linguistics. Especially I expect a progress of machine learning to represent a complicated model of language, perception and concept. The existing machine-based models are too simple to explain human language and cognition. Just think about bayesian network, perceptron, decision tree, Kernel approaches for clustering and classification, etc. The point could be how to deal with 'structural information' (like graph, sequence, hierarchy) and its change in models, in addition to attributes (or states). I expect information science and technology to provide smart tools to handle these. The another difficulty is still how to assess the semantic modeling provied cognitive linguistics scholars. Therefore, we need innovation in cognitive science, too. One of most important project is building a computational model of metapher and analogy, and the breakthrough should appear from this field. We need to come back to Gentner or Holyoak 's initial works and give them more firm basis from informational perspectives.