I am comparing the vocabulary of two sign languages in two locations to see if the vocabulary use in statistically different between locations. The data below is a sample of three words in the 250 word data set. The numbers in the synonym columns are the frequencies of use by Deaf subjects. The number of synonyms signed varies within words and study sites. So far I have concluded I can do a fisher's exact test with each word or I can do a Chi-squared test of the frequency of the same signed used in each test site compared to the expected for all the words in the data. My question is can I average the p-values form the fisher exact test for each word or should I use the Chi- squared test for the whole data set, or is there a better stats test to use?


Word Language Syn1 Syn2 Syn3 Syn4 Syn5
yellow 1 1 1 4 0
2 0 1 1 2

angry 1 2 5 0
2 1 1 3

Love 1 1 3 3 0 0
2 0 1 2 1 1
"syn" = synonym