Phrase usage frequency sources - what do you use, if any?

I would like to ask everyone: Are there are any methods / sources that you favor for investigating the usage frequency of words and phrases?

I used to simply go by basic Google searches, but it was recently brought to my attention that Google's estimate of search hits as displayed just below the search bar to the left, is wildly inaccurate, sometimes by an order of 4.

Additionally, some pages online are created from templates based on users' searches, and Google will also by default count occurrences that are from pages that exist in duplicate.

For now I'm thinking to decide on one of the many available text corpora available online. Here's a list of some of those:
http://view.byu.edu/

While these corpora reflect actual real-world usage, which would you prefer: Web-based corpora, or book-based corpora such as Google's ngrams?

Or perhaps, is it better to rely on smaller collections of example sentences that have been peer-reviewed by official or de-facto authorities on the English language?

While any native speaker can produce a vast number of statements, there will always be a number of people speaking the same language that would consider a number of those sentenced incorrect/ungrammatical/etc.

Should a native speaker creating sentences aim to create sentences that are likely to be accepted as 100% correct by as many individuals as possible?

What I did to reduce the hassle was,

I have several pages of sentences I need to open and look at... To illustrate:

-I open page 1 in a new browser window
-I open every sentence in page 1 in separate tabs
-I do my work for each individual sentence
-When I have finished the last entries, the first ones should be safe to close down - so I check whether the site finished processing.

Though this may not be feasible depending on your computer...

How about other options like:
1. Accepting donations.
2. Finding institutions willing to host the project - e.g. universities that may have an academic interest in supporting this project?

I'm having no trouble adding sentences:
http://93.20.168.172/eng/contributions/latest

However, I find that the language auto-detect doesn't work for me when adding sentences.

And we're back! Thanks for working it all out Trang.

I think we would all like to see as many audio recordings in as many languages as possible.

http://tatoeba.org/eng/stats/sentences_by_language
Here you see that audio recordings are mainly English.

We certainly need more Chinese contributors on tatoeba in general. If you know any, tip them about the site :)

I think the few active Chinese contributors here would be pleased to know that people are requesting audio recordings.

I don't believe this is such an option. The search function is very basic at this point.