Anastasios Asimakopoulos

    Hello @karl-hannay. Raw frequency, or just frequency, is how many times your search term occurs in the corpus, while relative frequency, aka normalised, is how many times your term occurs per x number of tokens, usually per 10K, 100K or 1 million tokens (depends on the corpus tool). Relative frequency allows us to compare the occurrences of a word between corpora of different sizes. Let me give you an example. The lemma idea occurs 4,003 times in the BAWE corpus and 1,607 times in the BASE corpus (British Academic Spoken English). We wouldn’t be able to say that idea is more frequent in BAWE than in BASE  because the two corpora are not the same size; BAWE is about 8,336,262 tokens but BASE is 1,756,545 tokens. So, looking at the relative frequencies of idea in the two corpora, we can say that it is more frequent is BASE than in BAWE: 914.86 per million tokens in BASE vs. 480.19 per million tokens in BAWE. I hope this helps.