You could create a function and use that. You could even combine it with one of the purrr::map() functions.
Otherwise, for the last bit:
word_year <- word_year %>% mutate(sev_word=(sumVAprod_word/sum_repet_word), # V+A
aro_word=(sumAprod_word/sum_repet_word), # A
val_word=(sumVprod_word/sum_repet_word)) # V
Anyway, a reproducible example of df_word2 and word_year would help.