group_modify does not behave as expected when the function is psych::alpha()

I'm trying to run psych::alpha on a dataset grouped by country. group_map works but as expected the list doesn't say countries, it indexes the countries ([[1]] etc) .
The reference site doesn't imply there needs to be any argument additions between group_map and group_modify
Passing through group_modify gives me the error:

Number of categories should be increased  in order to count frequencies. 
Error: The result of .f should be a data frame.
 1. `%>%`(...)
 3. dplyr:::group_modify.grouped_df(., ~psych::alpha(.x, check.keys = TRUE))
 5., fun, .keep = .keep)
 6. dplyr:::map2(chunks, group_keys, .f, ...)
 7. base::mapply(.f, .x, .y, MoreArgs = list(...), SIMPLIFY = FALSE)

This happens in my dataset where I

df%>% select(groupVar, vars1:var4)%>%group_by(groupVar)%>%group_modify(~ psych::alpha(.x, check.keys = TRUE))

and with example code

iris %>% group_by(Species) %>% 
  group_modify(~ psych::alpha(.x, check.keys = TRUE))

PS to mods: group_map is missing from packages tag

Adding link to associated GitHub issue