There are a number of repositories of linguistic resources, but they are either proprietary or they make the resources available at a low price for research purposes only, while the conditions for commercial use are very vague (write us). The low price hardly covers the cost of distribution and certainly is not enough to cover maintenance. The gold wagon is expected to come in from industry.
As a consequence, in building NLP applications many industrial corporations prefer to develop their own resources from scratch rather than being dependent on others. Research whose results might become economically interesting can not be based on such resources.
In fact, the situation is remarkably like that for software in the eighties, and the same solution should be considered:
Basic linguistic resources like grammars, lexica, parsers, corpora and ontologies should be made freely available in the public domain, especially if they have been developed with public money. Their users should be invited to contribute improvements, thus enabling a low-cost form of maintenance.Where have we heard this before?