In a database of 18 major newspapers, the relative frequency of phrases like "race prejudice" has declined by 60percentt since 1980 (as far back as the online records go), while the frequency of "racism" has doubled.
Using the random model expected three-phrase combination frequencies were calculated as the product of the frequency of each phrase.
The much lower frequency of this phrase in regard to President Bush implies that Bush is first in line, is the legitimate heir to the nomination of his party, whereas Kerry for some reason is not.
And, as you can see from this Google chart that tracks the frequency of a phrase over time, "running the gauntlet" (shown in red) has been the dominant usage for a very long time.
The horizontal axis represents the actual frequency of the phrase, and the vertical axis represents the counts of the returning results.
This allows users to scour the archive of full-text searchable historical documents to see the frequency of specific phrases.
We tested whether the frequency of three-phrase combinations deviated from two models: a simple random model and a 1st order Markov model, the first-order model (see Methods).
We compared the frequency of three-phrase combinations to the first-order model since in the previous analysis it was the more complex model and was a closer fit to the data.
This study was focused on encoding of low repetition rates (1, 6, 10, 14, 22, and 30 Hz), a range associated with the occurrence frequency of vocalization phrases, phonemes, or syllables (gray background in Figs. 1B, S1C, S1D) and with a high probability of encountering high temporal response fidelity.
